Oracle SQL Interview Questions — APEX Interview Mastery

SQL Basics & Commands

Q1. What is SQL?

EMPLOYEE_ID	EMPLOYEE_NAME	SALARY
104	Dave Brown	70,000
102	Bob Smith	62,000
103	Carol White	55,000

Q2. What are the different types of SQL commands?

CATEGORY	COMMANDS	EXAMPLE
DDL	CREATE, ALTER, DROP, TRUNCATE	CREATE TABLE employees(...)
DML	INSERT, UPDATE, DELETE, MERGE	UPDATE employees SET salary=60000
DCL	GRANT, REVOKE	GRANT SELECT ON employees TO user1
TCL	COMMIT, ROLLBACK, SAVEPOINT	COMMIT

Q3. What is the SELECT statement and what is the order of its clauses?

EMP_NAME	SALARY
Carol White	55,000
Alice Johnson	50,000

Q4. What does the DISTINCT keyword do in SQL?

DEPT_ID
10
20

Keys & Constraints

Q5. What is a primary key?

Q6. What is a foreign key?

Q7. What is a unique key?

EMP_ID	EMAIL	STATUS
101	alice@company.com	✅ Inserted OK
102	alice@company.com	❌ ORA-00001: unique constraint (emp_email) violated
103	carol@company.com	✅ Inserted OK — different email

Q8. Discuss the role of constraints in database design.

Q9. Explain the difference between a PRIMARY KEY and a UNIQUE constraint in Oracle.

Answer

Both enforce uniqueness, but a table can have only one PRIMARY KEY while it can have multiple UNIQUE constraints.

PRIMARY KEY automatically creates a unique index and cannot contain NULL values.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

CodeOracle SQL

SQL

-- PRIMARY KEY characteristics
CREATE TABLE employees_pk_demo (
    employee_id NUMBER PRIMARY KEY,  -- Only one PRIMARY KEY per table
    ssn NUMBER UNIQUE,               -- Multiple UNIQUE constraints allowed
    email VARCHAR2(100) UNIQUE,
    phone VARCHAR2(20) UNIQUE,
    first_name VARCHAR2(50) NOT NULL,
    last_name VARCHAR2(50) NOT NULL
);

-- PRIMARY KEY cannot be NULL
-- This will fail:
-- INSERT INTO employees_pk_demo VALUES (NULL, '123456789', 'john@email.com', '555-1234', 'John', 'Doe');

-- UNIQUE constraint can have NULL values (but only one NULL per column)
INSERT INTO employees_pk_demo VALUES (1, NULL, 'john@email.com', '555-1234', 'John', 'Doe');
INSERT INTO employees_pk_demo VALUES (2, NULL, 'jane@email.com', '555-5678', 'Jane', 'Smith');

-- Composite PRIMARY KEY
CREATE TABLE order_items (
    order_id NUMBER,
    product_id NUMBER,
    quantity NUMBER,
    unit_price NUMBER,
    PRIMARY KEY (order_id, product_id)  -- Composite primary key
);

-- Composite UNIQUE constraint
CREATE TABLE employees_unique_demo (
    employee_id NUMBER PRIMARY KEY,
    first_name VARCHAR2(50),
    last_name VARCHAR2(50),
    department_id NUMBER,
    UNIQUE (first_name, last_name, department_id)  -- Composite unique constraint
);

-- Adding constraints after table creation
ALTER TABLE employees_pk_demo 
ADD CONSTRAINT uk_emp_ssn UNIQUE (ssn);

-- Check constraint information
SELECT constraint_name, constraint_type, search_condition, status
FROM user_constraints
WHERE table_name = 'EMPLOYEES_PK_DEMO';

-- Index creation behavior
-- PRIMARY KEY automatically creates a unique index
-- UNIQUE constraint also creates a unique index

SELECT index_name, table_name, uniqueness, column_name
FROM user_ind_columns
WHERE table_name = 'EMPLOYEES_PK_DEMO'
ORDER BY index_name, column_position;

-- Differences summary:
-- 1. PRIMARY KEY: One per table, cannot be NULL, automatically indexed
-- 2. UNIQUE: Multiple per table, can have one NULL value, automatically indexed
-- 3. Both prevent duplicate values
-- 4. PRIMARY KEY is used for referential integrity (foreign keys)
-- 5. Both support composite (multi-column) constraints

-- Foreign key references PRIMARY KEY
CREATE TABLE departments (
    department_id NUMBER PRIMARY KEY,
    department_name VARCHAR2(100) UNIQUE NOT NULL
);

CREATE TABLE employees_fk_demo (
    employee_id NUMBER PRIMARY KEY,
    department_id NUMBER,
    FOREIGN KEY (department_id) REFERENCES departments(department_id)
);

-- Foreign key can also reference UNIQUE constraint
CREATE TABLE employees_fk_unique (
    employee_id NUMBER PRIMARY KEY,
    department_name VARCHAR2(100),
    FOREIGN KEY (department_name) REFERENCES departments(department_name)
);

-- Performance comparison
-- Both PRIMARY KEY and UNIQUE constraints use B-tree indexes
-- Performance is similar for lookups and joins
-- PRIMARY KEY is slightly faster for foreign key lookups

-- Disable/Enable constraints
ALTER TABLE employees_pk_demo DISABLE CONSTRAINT uk_emp_ssn;
ALTER TABLE employees_pk_demo ENABLE CONSTRAINT uk_emp_ssn;

-- Drop constraints
ALTER TABLE employees_pk_demo DROP CONSTRAINT uk_emp_ssn;

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

Data state before query executes.

Output / Afterresult

After

EMP_ID	EMP_NAME	DEPT_ID	SALARY
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	20	70,000

All 4 rows returned — ordered by salary ASC

Key Takeaway

Both enforce uniqueness, but a table can have only one PRIMARY KEY while it can have multiple UNIQUE constraints.

Q10. What is a NOT NULL constraint in Oracle?

ERROR_CODE	MESSAGE
ORA-01400	cannot insert NULL into ("EMPLOYEE_NAME")

Q11. What is a CHECK constraint in Oracle?

ERROR_CODE	MESSAGE
ORA-02290	check constraint violated

Q12. What is a DEFAULT constraint in Oracle?

ORDER_ID	ORDER_DATE	STATUS
1001	(today's date)	PENDING

Q13. What is a composite key and when would you use one?

student_id	course_id	enrolled_date
1	101	2026-01-10
1	102	2026-01-12
2	101	2026-01-15

CONSTRAINT_NAME	CONSTRAINT_TYPE	COLUMNS
PK_ENROLLMENTS	P	STUDENT_ID, COURSE_ID

Joins

Q14. Explain the concept of a join in SQL.

Q15. What are the different types of joins?

Answer

In SQL, different types of joins enable the combination of rows from two or more tables.

The most common type is the INNER JOIN, which returns rows when there is a match in both tables.

The LEFT JOIN, also known as LEFT OUTER JOIN, returns all rows from the left table and the matched rows from the right table, filling in with NULLs if there is no match.

Similarly, the RIGHT JOIN, or RIGHT OUTER JOIN, includes all rows from the right table and the matched rows from the left table, using NULLs where no match exists.

The FULL OUTER JOIN combines the results of both LEFT and RIGHT JOINS, displaying all rows from both tables with NULLs in places where there is no match.

Another type, the CROSS JOIN, produces a Cartesian product of the two tables, joining every row of the first table with every row of the second table.

Lastly, the SELF JOIN is a regular join but the table is joined with itself.

Sample Tablesused in all examples

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

INNER JOINMatching rows only

Before — INNER JOINinitial state

Before INNER JOIN

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

CodeOracle SQL

SQL

SELECT e.emp_name, d.dept_name
FROM   employees   e
INNER JOIN departments d ON e.dept_id = d.dept_id;
-- Only rows matching on BOTH sides

After / Result3 rows

After

EMP_NAME	DEPT_NAME
Alice Johnson	Engineering
Bob Smith	Marketing
Carol White	Engineering

3 rows — only rows with matching dept_id in both tables

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

CodeOracle SQL

SQL

SELECT e.emp_name, d.dept_name
FROM   employees   e
LEFT JOIN departments d ON e.dept_id = d.dept_id;
-- ALL employees returned

After / Result4 rows

After — LEFT JOIN result

EMP_NAME	DEPT_NAME
Alice Johnson	Engineering
Bob Smith	Marketing
Carol White	Engineering
Dave Brown	NULL

4 rows — Dave included with NULL dept · Finance excluded

RIGHT JOINAll right rows + NULL where no match

Before — RIGHT JOINinitial state

Before RIGHT JOIN

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

CodeOracle SQL

SQL

SELECT e.emp_name, d.dept_name
FROM   employees   e
RIGHT JOIN departments d ON e.dept_id = d.dept_id;
-- ALL departments returned

After / Result4 rows

After — RIGHT JOIN result

EMP_NAME	DEPT_NAME
Alice Johnson	Engineering
Carol White	Engineering
Bob Smith	Marketing
NULL	Finance

4 rows — Finance included with NULL emp · Dave excluded

FULL OUTER JOINAll rows from both tables

Before — FULL OUTER JOINinitial state

Before FULL OUTER JOIN

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

CodeOracle SQL

SQL

SELECT e.emp_name, d.dept_name
FROM   employees   e
FULL OUTER JOIN departments d ON e.dept_id = d.dept_id;
-- ALL rows both sides

After / Result5 rows

After — FULL OUTER JOIN result

EMP_NAME	DEPT_NAME
Alice Johnson	Engineering
Bob Smith	Marketing
Carol White	Engineering
Dave Brown	NULL
NULL	Finance

5 rows — everyone included · NULLs where no match

Quick Comparisonall 4 join types

Join Type	Dave?	Finance?	Rows
INNER JOIN	❌ No	❌ No	3
LEFT JOIN	✅ Yes (NULL dept)	❌ No	4
RIGHT JOIN	❌ No	✅ Yes (NULL emp)	4
FULL OUTER JOIN	✅ Yes	✅ Yes	5

Key Takeaway

In SQL, different types of joins enable the combination of rows from two or more tables.

Q16. What is the difference between INNER JOIN and OUTER JOIN?

Q17. What is a self-join and provide an example of its use.

Q18. Explain the use of lateral joins and provide a use case.

Q19. Explain JOIN and different types of JOIN in Oracle SQL.

Answer

JOIN is used to combine rows from two or more tables based on a related column between them.

Oracle supports multiple JOIN types.

Sample Tablesused in all examples

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

INNER JOINMatching rows only

Before — INNER JOINinitial state

Before INNER JOIN

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

CodeOracle SQL

SQL

SELECT e.emp_name, d.dept_name
FROM   employees   e
INNER JOIN departments d ON e.dept_id = d.dept_id;
-- Only rows matching on BOTH sides

After / Result3 rows

After

EMP_NAME	DEPT_NAME
Alice Johnson	Engineering
Bob Smith	Marketing
Carol White	Engineering

3 rows — only rows with matching dept_id in both tables

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

CodeOracle SQL

SQL

SELECT e.emp_name, d.dept_name
FROM   employees   e
LEFT JOIN departments d ON e.dept_id = d.dept_id;
-- ALL employees returned

After / Result4 rows

After — LEFT JOIN result

EMP_NAME	DEPT_NAME
Alice Johnson	Engineering
Bob Smith	Marketing
Carol White	Engineering
Dave Brown	NULL

4 rows — Dave included with NULL dept · Finance excluded

RIGHT JOINAll right rows + NULL where no match

Before — RIGHT JOINinitial state

Before RIGHT JOIN

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

CodeOracle SQL

SQL

SELECT e.emp_name, d.dept_name
FROM   employees   e
RIGHT JOIN departments d ON e.dept_id = d.dept_id;
-- ALL departments returned

After / Result4 rows

After — RIGHT JOIN result

EMP_NAME	DEPT_NAME
Alice Johnson	Engineering
Carol White	Engineering
Bob Smith	Marketing
NULL	Finance

4 rows — Finance included with NULL emp · Dave excluded

FULL OUTER JOINAll rows from both tables

Before — FULL OUTER JOINinitial state

Before FULL OUTER JOIN

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

CodeOracle SQL

SQL

SELECT e.emp_name, d.dept_name
FROM   employees   e
FULL OUTER JOIN departments d ON e.dept_id = d.dept_id;
-- ALL rows both sides

After / Result5 rows

After — FULL OUTER JOIN result

EMP_NAME	DEPT_NAME
Alice Johnson	Engineering
Bob Smith	Marketing
Carol White	Engineering
Dave Brown	NULL
NULL	Finance

5 rows — everyone included · NULLs where no match

Quick Comparisonall 4 join types

Join Type	Dave?	Finance?	Rows
INNER JOIN	❌ No	❌ No	3
LEFT JOIN	✅ Yes (NULL dept)	❌ No	4
RIGHT JOIN	❌ No	✅ Yes (NULL emp)	4
FULL OUTER JOIN	✅ Yes	✅ Yes	5

Key Takeaway

JOIN is used to combine rows from two or more tables based on a related column between them.

Q20. What is a CROSS JOIN and how does it differ from other joins?

Q21. What is a FULL OUTER JOIN and when is it used?

Filtering, Grouping & Aggregates

Q22. Explain the use of the GROUP BY clause.

Q23. What is the difference between HAVING and WHERE clauses?

Q24. What are aggregate functions in SQL?

Q25. How do IN, BETWEEN, and LIKE operators work for filtering data?

Q26. What do ROLLUP and CUBE do in a GROUP BY clause?

Subqueries & CTEs

Q27. What is a subquery in SQL?

Q28. What are Common Table Expressions (CTEs) and when would you use them?

Answer

Common Table Expressions (CTEs) are temporary result sets that simplify complex joins and subqueries in SQL.

CTEs provide a way to create a temporary named result set, which is available only within the execution scope of a single SELECT, INSERT, UPDATE, or DELETE statement.

CTEs improve readability and maintenance of complex queries by breaking them into simpler, reusable parts.

You would use CTEs in situations where you need to reference the result set multiple times within a single query, or when dealing with hierarchical data structures.

They are particularly useful in recursive queries, which are essential for dealing with hierarchical or tree-structured data.

CTEs offer a more readable and organized approach compared to derived tables or subqueries, especially in queries that involve multiple levels of data extraction and manipulation.

They ensure better performance and readability in complex SQL queries, making them a preferred choice for SQL developers and database administrators.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

CodeOracle SQL

SQL

WITH dept_avg AS (
  SELECT department_id, AVG(salary) AS avg_sal
  FROM   employees GROUP BY department_id
)
SELECT e.employee_name, e.salary
FROM   employees e JOIN dept_avg d ON e.department_id = d.department_id
WHERE  e.salary > d.avg_sal;

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

Data state before query executes.

Output / Afterresult

After

EMP_NAME	DEPT_NAME
Alice Johnson	Engineering
Bob Smith	Marketing
Carol White	Engineering

3 rows — only rows with matching dept_id in both tables (Dave Brown excluded — no dept)

Key Takeaway

Common Table Expressions (CTEs) are temporary result sets that simplify complex joins and subqueries in SQL.

Q29. What are correlated subqueries and how do they differ from regular subqueries?

Q30. Discuss the use of recursive queries in SQL.

Q31. Explain the concept of 'with clause' in SQL.

Answer

The 'WITH clause' in SQL, also known as a Common Table Expression (CTE), is a powerful feature that allows you to define temporary result sets within a SQL query.

This temporary result set is referenced within the main query, making complex SQL queries more manageable and readable.You start by specifying a CTE name and then define the query that generates the result set.

This result set includes filtering, joining, or aggregating data as needed.

Reference this CTE within the main query once defined, as if it were a table or subquery.The benefits of using the 'WITH clause' are twofold.

It enhances the readability of SQL queries by breaking them down into logical, named parts.

This makes it easier for developers to understand and maintain complex queries.

It also improves query performance since the database engine optimizes the execution plan based on the CTE definition.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

CodeOracle SQL

SQL

WITH dept_avg AS (
  SELECT department_id, AVG(salary) AS avg_sal
  FROM   employees GROUP BY department_id
)
SELECT e.employee_name, e.salary
FROM   employees e JOIN dept_avg d ON e.department_id = d.department_id
WHERE  e.salary > d.avg_sal;

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

Data state before query executes.

Output / Afterresult

After

EMP_NAME	DEPT_NAME
Alice Johnson	Engineering
Bob Smith	Marketing
Carol White	Engineering

3 rows — only rows with matching dept_id in both tables (Dave Brown excluded — no dept)

Key Takeaway

The 'WITH clause' in SQL, also known as a Common Table Expression (CTE), is a powerful feature that allows you to define temporary result sets within a SQL query.

Q32. Write a query to find employees who earn more than their manager using correlated subqueries.

Answer

Use correlated subqueries or self-joins to compare employee salaries with their manager's salaries.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

CodeOracle SQL

SQL

-- Method 1: Using correlated subquery
SELECT e.employee_id, e.first_name, e.last_name, e.salary, e.manager_id
FROM employees e
WHERE e.salary > (
   SELECT m.salary
   FROM employees m
   WHERE m.employee_id = e.manager_id
);

-- Method 2: Using self-join
SELECT e.employee_id, e.first_name, e.last_name, e.salary as emp_salary,
      m.employee_id as mgr_id, m.first_name as mgr_name, m.salary as mgr_salary
FROM employees e
JOIN employees m ON e.manager_id = m.employee_id
WHERE e.salary > m.salary;

-- Method 3: Using window functions (more efficient for large datasets)
SELECT employee_id, first_name, last_name, salary, manager_salary
FROM (
   SELECT employee_id, first_name, last_name, salary, manager_id,
          LAG(salary) OVER (PARTITION BY manager_id ORDER BY employee_id) as manager_salary
   FROM employees
)
WHERE salary > manager_salary;

-- Method 4: Include additional analysis
SELECT e.employee_id, e.first_name, e.salary as emp_salary,
      m.salary as mgr_salary,
      e.salary - m.salary as salary_difference,
      ROUND((e.salary - m.salary) / m.salary * 100, 2) as pct_higher
FROM employees e
JOIN employees m ON e.manager_id = m.employee_id
WHERE e.salary > m.salary
ORDER BY salary_difference DESC;

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

Data state before query executes.

Output / Afterresult

After

EMP_NAME	SALARY	DEPT_AVG
Dave Brown	70,000	66,000
Bob Smith	62,000	66,000

2 rows — employees earning above their own department average

Key Takeaway

Use correlated subqueries or self-joins to compare employee salaries with their manager's salaries.

Q33. What is the difference between EXISTS, IN, and a subquery for filtering?

Set Operators

Q34. Explain the difference between UNION and UNION ALL commands.

Q35. What are the differences between UNION and UNION ALL in Oracle?

Answer

UNION combines the result sets of two or more queries and removes duplicates.

UNION ALL combines result sets but does not remove duplicates.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

CodeOracle SQL

SQL

-- UNION: Removes duplicates and sorts results (slower)
SELECT employee_id, first_name FROM employees WHERE department_id = 10
UNION
SELECT employee_id, first_name FROM employees WHERE department_id = 20;

-- UNION ALL: Keeps all rows including duplicates (faster)
SELECT employee_id, first_name FROM employees WHERE department_id = 10
UNION ALL
SELECT employee_id, first_name FROM employees WHERE department_id = 20;

-- Performance comparison
-- UNION performs sorting and duplicate removal
-- UNION ALL is faster as it doesn't sort or remove duplicates

-- Example with different data types (must be compatible)
SELECT employee_id, first_name, 'EMPLOYEE' as record_type FROM employees
UNION ALL
SELECT department_id, department_name, 'DEPARTMENT' as record_type FROM departments;

-- Real-world example: Combining current and archived data
SELECT order_id, customer_id, order_date, 'ACTIVE' as status
FROM current_orders
UNION ALL
SELECT order_id, customer_id, order_date, 'ARCHIVED' as status
FROM archived_orders
ORDER BY order_date DESC;

-- When to use UNION vs UNION ALL:
-- Use UNION when you need unique records across sets
-- Use UNION ALL when you want all records including duplicates
-- UNION ALL is generally preferred for performance unless duplicates must be removed

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

Data state before query executes.

Output / Afterresult

After

EMP_ID	EMP_NAME	SOURCE
101	Alice Johnson	CURRENT
102	Bob Smith	CURRENT
101	Alice Johnson	ARCHIVE
102	Bob Smith	ARCHIVE

4 rows — UNION ALL keeps duplicates, faster than UNION

Key Takeaway

UNION combines the result sets of two or more queries and removes duplicates.

Q36. What do INTERSECT and MINUS do in Oracle SQL?

Views & Materialized Views

Q37. What is a view in SQL?

Q38. What are materialized views and how do they differ from standard views?

Answer

Materialized views are database objects in SQL that store the results of a query as a physical table.

Materialized views precompute and store the data, making them faster to retrieve, unlike standard views, which are virtual and execute the underlying query each time they are accessed.The key difference between materialized views and standard views is that materialized views contain precomputed data, while standard views are virtual representations of data from underlying tables.

The database engine executes the underlying query in real-time to retrieve the data, when you query a standard view.

A materialized view stores the data in a physical table, so querying it is faster because it avoids the need for real-time computation.Materialized views offer benefits in terms of query performance optimization, especially when dealing with complex and resource-intensive queries.

They reduce the overhead of recalculating the same result set repeatedly by maintaining a snapshot of the data.It's important to note that materialized views come with trade-offs.

They consume storage space and need to be refreshed periodically to ensure that the stored data remains up-to-date with the source data.

This refresh process introduces some latency in accessing the latest data.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

CodeOracle SQL

SQL

CREATE OR REPLACE VIEW v_dept_employees AS
SELECT e.employee_name, e.salary, d.dept_name
FROM   employees e JOIN departments d ON e.dept_id = d.dept_id
WHERE  e.salary > 40000;

SELECT * FROM v_dept_employees ORDER BY salary DESC;

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

Data state before query executes.

Output / Afterresult

After

EMP_NAME	DEPT_NAME
Alice Johnson	Engineering
Bob Smith	Marketing
Carol White	Engineering

3 rows — only rows with matching dept_id in both tables (Dave Brown excluded — no dept)

Key Takeaway

Materialized views are database objects in SQL that store the results of a query as a physical table.

Q39. What is a view in Oracle and how do you create one?

Answer

A view is a virtual table based on the result set of an SQL statement.

It contains rows and columns from one or more tables but doesn't store data physically.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

CodeOracle SQL

SQL

-- 1. Simple View
CREATE VIEW v_employee_basic AS
SELECT employee_id, first_name, last_name, email, hire_date
FROM employees;

-- Query the view
SELECT * FROM v_employee_basic WHERE employee_id = 101;

-- 2. Complex View with Joins
CREATE VIEW v_employee_details AS
SELECT e.employee_id,
       e.first_name,
       e.last_name,
       e.email,
       e.salary,
       d.department_name,
       j.job_title,
       m.first_name || ' ' || m.last_name as manager_name
FROM employees e
LEFT JOIN departments d ON e.department_id = d.department_id
LEFT JOIN jobs j ON e.job_id = j.job_id
LEFT JOIN employees m ON e.manager_id = m.employee_id;

-- 3. View with Calculations
CREATE VIEW v_employee_summary AS
SELECT department_id,
       COUNT(*) as employee_count,
       AVG(salary) as avg_salary,
       MIN(salary) as min_salary,
       MAX(salary) as max_salary,
       SUM(salary) as total_salary
FROM employees
GROUP BY department_id;

-- 4. Updatable View
CREATE VIEW v_emp_updatable AS
SELECT employee_id, first_name, last_name, email, salary
FROM employees
WHERE department_id = 10
WITH CHECK OPTION;

-- Update through view
UPDATE v_emp_updatable 
SET salary = salary * 1.1 
WHERE employee_id = 101;

-- 5. View with Column Aliases
CREATE VIEW v_emp_formatted (emp_id, full_name, annual_salary, hire_year) AS
SELECT employee_id,
       first_name || ' ' || last_name,
       salary * 12,
       EXTRACT(YEAR FROM hire_date)
FROM employees;

-- 6. Materialized View (for performance)
CREATE MATERIALIZED VIEW mv_dept_summary
BUILD IMMEDIATE
REFRESH FAST ON COMMIT
AS
SELECT d.department_id,
       d.department_name,
       COUNT(e.employee_id) as emp_count,
       AVG(e.salary) as avg_salary
FROM departments d
LEFT JOIN employees e ON d.department_id = e.department_id
GROUP BY d.department_id, d.department_name;

-- 7. View with Security (Row-Level Security)
CREATE VIEW v_emp_security AS
SELECT employee_id, first_name, last_name, salary
FROM employees
WHERE department_id = (SELECT department_id 
                       FROM employees 
                       WHERE employee_id = SYS_CONTEXT('APEX$SESSION', 'APP_USER'));

-- 8. Instead-Of Trigger for Complex Views
CREATE OR REPLACE VIEW v_emp_dept_complex AS
SELECT e.employee_id,
       e.first_name,
       e.last_name,
       d.department_name
FROM employees e
JOIN departments d ON e.department_id = d.department_id;

CREATE OR REPLACE TRIGGER trg_v_emp_dept_insert
INSTEAD OF INSERT ON v_emp_dept_complex
FOR EACH ROW
BEGIN
    INSERT INTO employees (employee_id, first_name, last_name, department_id)
    VALUES (:NEW.employee_id, :NEW.first_name, :NEW.last_name,
            (SELECT department_id FROM departments WHERE department_name = :NEW.department_name));
END;
/

-- View Management
-- Drop view
DROP VIEW v_employee_basic;

-- Replace view
CREATE OR REPLACE VIEW v_employee_basic AS
SELECT employee_id, first_name, last_name, email, phone_number
FROM employees;

-- View metadata
SELECT view_name, text
FROM user_views
WHERE view_name LIKE 'V_EMP%';

-- Check if view is updatable
SELECT table_name, column_name, updatable, insertable, deletable
FROM user_updatable_columns
WHERE table_name = 'V_EMP_UPDATABLE';

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

Data state before query executes.

Output / Afterresult

After

EMPLOYEE	MANAGER
Bob Smith	Alice Johnson
Carol White	Alice Johnson
Dave Brown	Bob Smith

3 rows — self join pairs each employee with their manager's name

Key Takeaway

A view is a virtual table based on the result set of an SQL statement.

Indexes & Indexing Strategy

Q40. What is an index in SQL?

Q41. Explain the difference between clustered and non-clustered indexes.

Q42. How do you implement indexing in a large database and what factors do you consider?

Answer

Implementing indexing in a large database involves creating indexes on columns that are frequently used in search conditions, JOIN operations, or as part of WHERE clauses.

Indexing significantly enhances query performance by reducing the amount of data the database engine needs to scan.

Consider the frequency of the query types and the specific columns involved in these queries, when determining which columns to index.

Opt for composite indexes if multiple columns are often used together in queries.Ensure the database maintains index efficiency by periodically reviewing and updating the indexes based on query patterns and data changes.

Avoid over-indexing as it leads to increased storage requirements and slower data modification operations.

Index maintenance is crucial, especially in dynamic databases where data insertion, updating, and deletion are frequent.

Implement indexing carefully in large databases, ensuring it aligns with the database's usage patterns and query requirements.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

CodeOracle SQL

SQL

-- Create index
CREATE INDEX idx_emp_dept ON employees(department_id);

-- Check execution plan
EXPLAIN PLAN FOR
  SELECT * FROM employees WHERE department_id = 10;
SELECT * FROM TABLE(DBMS_XPLAN.DISPLAY);

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

Data state before query executes.

Output / Afterresult

After

EMP_ID	EMP_NAME	DEPT_ID	SALARY
101	Alice Johnson	10	50,000
103	Carol White	10	55,000

2 rows — Bob Smith, Dave Brown excluded by WHERE clause

Key Takeaway

Implementing indexing in a large database involves creating indexes on columns that are frequently used in search conditions, JOIN operations, or as part of WHERE clauses.

Q43. Discuss advanced indexing techniques like bitmap indexes and partial indexes in SQL.

Q44. How do you create an index on a column in Oracle?

Answer

Use the CREATE INDEX statement to create an index on one or more columns to improve query performance.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

CodeOracle SQL

SQL

-- Create a simple index on a single column
CREATE INDEX idx_employee_last_name ON employees(last_name);

-- Create a composite index on multiple columns
CREATE INDEX idx_emp_dept_salary ON employees(department_id, salary);

-- Create a unique index
CREATE UNIQUE INDEX idx_employee_email ON employees(email);

-- Create a function-based index
CREATE INDEX idx_emp_upper_name ON employees(UPPER(last_name));

-- Create a partial index with WHERE clause (Oracle 12c+)
CREATE INDEX idx_active_employees ON employees(employee_id) 
WHERE status = 'ACTIVE';

-- Create index with specific options
CREATE INDEX idx_emp_hire_date ON employees(hire_date)
TABLESPACE users
PCTFREE 10
STORAGE (INITIAL 1M NEXT 1M);

-- Check index usage
SELECT index_name, table_name, column_name, column_position
FROM user_ind_columns
WHERE table_name = 'EMPLOYEES'
ORDER BY index_name, column_position;

-- Monitor index usage
SELECT index_name, table_name, monitoring, used
FROM v$object_usage
WHERE table_name = 'EMPLOYEES';

-- Enable monitoring for an index
ALTER INDEX idx_employee_last_name MONITORING USAGE;

-- Drop an index
DROP INDEX idx_employee_last_name;

-- Rebuild an index
ALTER INDEX idx_emp_dept_salary REBUILD;

-- Create bitmap index (for low cardinality columns)
CREATE BITMAP INDEX idx_emp_gender ON employees(gender);

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

Data state before query executes.

Output / Afterresult

After

EMP_ID	EMP_NAME	DEPT_ID	SALARY
101	Alice Johnson	10	50,000
103	Carol White	10	55,000
102	Bob Smith	20	62,000
104	Dave Brown	20	70,000

All 4 rows returned — ordered by salary ASC

Key Takeaway

Use the CREATE INDEX statement to create an index on one or more columns to improve query performance.

Q45. What is a composite (multi-column) index and when is it useful?

Q46. What is a function-based index in Oracle?

DDL & Schema Management

Q47. Explain the difference between DELETE and TRUNCATE commands.

Answer

The difference between DELETE and TRUNCATE commands is that DELETE removes rows from a table based on a specified condition, while TRUNCATE is a Data Definition Language (DDL) command that quickly removes all rows from a table without logging the deletion of individual rows.

DELETE command logs individual row deletions, which makes it a slower process, especially in large tables.

DELETE operation consumes more transaction log space and permits the use of WHERE clause to specify which rows to remove.TRUNCATE resets the table to its empty state, providing a faster method for deleting all records, especially in large tables.

TRUNCATE also deallocates the data pages used by the table, resulting in the freeing of space for the database but It does not allow for conditional deletions as it does not support the WHERE clause.

Use TRUNCATE when the entire table needs to be emptied without the need for a transaction log for each row deletion.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

CodeOracle SQL

SQL

SELECT * FROM employees WHERE ROWNUM <= 5;

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

Data state before query executes.

Output / Afterresult

After

COMMAND	CONDITION	ROWS_AFFECTED	ROLLBACK	SPEED
DELETE FROM employees WHERE dept_id=10	dept=10	2 rows	✅ Yes	Slow
DELETE FROM employees	no filter	4 rows	✅ Yes	Slow
TRUNCATE TABLE employees	N/A	4 rows	❌ No	Fast

TRUNCATE is fastest but irreversible — DELETE is flexible and rollbackable

Key Takeaway

Q48. Explain the differences between DROP, DELETE, and TRUNCATE commands in Oracle with examples.

Answer

DROP removes the entire table structure and data, DELETE removes specific rows with WHERE conditions and can be rolled back, TRUNCATE removes all rows quickly but cannot be rolled back.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

CodeOracle SQL

SQL

-- CREATE sample table for demonstration
CREATE TABLE employees_demo (
    employee_id NUMBER PRIMARY KEY,
    first_name VARCHAR2(50),
    last_name VARCHAR2(50),
    salary NUMBER,
    department_id NUMBER
);

-- Insert sample data
INSERT ALL
    INTO employees_demo VALUES (1, 'John', 'Doe', 50000, 10)
    INTO employees_demo VALUES (2, 'Jane', 'Smith', 60000, 20)
    INTO employees_demo VALUES (3, 'Bob', 'Johnson', 55000, 10)
    INTO employees_demo VALUES (4, 'Alice', 'Brown', 65000, 30)
    INTO employees_demo VALUES (5, 'Tom', 'Wilson', 52000, 20)
SELECT * FROM dual;

-- Verify data
SELECT * FROM employees_demo;

-- ========================================
-- DELETE COMMAND
-- ========================================
-- Removes specific rows based on WHERE condition
-- Can also remove ALL rows if no WHERE clause is used
-- Can be rolled back
-- Slower for large datasets
-- Triggers fire for each deleted row

-- Delete specific employees
DELETE FROM employees_demo WHERE department_id = 10;
-- Removes employees with department_id = 10

-- Check remaining data
SELECT * FROM employees_demo;

-- Rollback is possible
ROLLBACK;

-- Data is restored
SELECT * FROM employees_demo;

-- Delete with more complex conditions
DELETE FROM employees_demo 
WHERE salary < 55000 AND department_id = 20;

-- Delete ALL rows from table (without WHERE clause)
DELETE FROM employees_demo;
-- This removes ALL data but keeps table structure
-- Same result as TRUNCATE but slower and can be rolled back

-- ========================================
-- TRUNCATE COMMAND  
-- ========================================
-- Removes ALL rows from table
-- Cannot be rolled back (DDL command)
-- Much faster than DELETE for large tables
-- Resets high water mark
-- Does not fire triggers

-- Restore data first
ROLLBACK;
INSERT ALL
    INTO employees_demo VALUES (1, 'John', 'Doe', 50000, 10)
    INTO employees_demo VALUES (2, 'Jane', 'Smith', 60000, 20)
    INTO employees_demo VALUES (3, 'Bob', 'Johnson', 55000, 10)
SELECT * FROM dual;

-- Truncate table (removes all data)
TRUNCATE TABLE employees_demo;

-- Check table - no data but structure exists
SELECT * FROM employees_demo;
DESC employees_demo;

-- Cannot rollback TRUNCATE
ROLLBACK; -- This won't restore data

-- ========================================
-- DROP COMMAND
-- ========================================
-- Removes entire table structure and data
-- Cannot be rolled back
-- Frees up storage space completely
-- Removes indexes, triggers, constraints

-- Recreate table with data
CREATE TABLE employees_demo (
    employee_id NUMBER PRIMARY KEY,
    first_name VARCHAR2(50),
    last_name VARCHAR2(50),
    salary NUMBER,
    department_id NUMBER
);

INSERT INTO employees_demo VALUES (1, 'John', 'Doe', 50000, 10);

-- Drop the entire table
DROP TABLE employees_demo;

-- Table no longer exists
-- SELECT * FROM employees_demo; -- This would cause error

-- Check if table exists
SELECT table_name FROM user_tables WHERE table_name = 'EMPLOYEES_DEMO';

-- ========================================
-- COMPARISON TABLE
-- ========================================

/*
+------------------+------------------+------------------+------------------+
| ASPECT           | DELETE           | TRUNCATE         | DROP             |
+------------------+------------------+------------------+------------------+
| Purpose          | Remove specific  | Remove all rows  | Remove entire    |
|                  | rows OR all rows |                  | table            |
+------------------+------------------+------------------+------------------+
| WHERE Clause     | Yes (optional)   | No               | No               |
|                  | Without WHERE =  |                  |                  |
|                  | deletes all rows |                  |                  |
+------------------+------------------+------------------+------------------+
| Rollback         | Yes (DML)        | No (DDL)         | No (DDL)         |
+------------------+------------------+------------------+------------------+
| Speed            | Slower           | Very Fast        | Fast             |
|                  | (even for all    |                  |                  |
|                  | rows)            |                  |                  |
+------------------+------------------+------------------+------------------+
| Table Structure  | Preserved        | Preserved        | Removed          |
+------------------+------------------+------------------+------------------+
| Triggers         | Fire             | Don't Fire       | Removed          |
+------------------+------------------+------------------+------------------+
| Indexes          | Preserved        | Preserved        | Removed          |
+------------------+------------------+------------------+------------------+
| Storage Space    | Not released     | Released         | Completely freed |
+------------------+------------------+------------------+------------------+
| Transaction Log  | Fully logged     | Minimally logged | Minimally logged |
+------------------+------------------+------------------+------------------+
| Identity/Sequence| Continues        | Resets           | Removed          |
+------------------+------------------+------------------+------------------+
*/

-- ========================================
-- PRACTICAL EXAMPLES
-- ========================================

-- Create table for examples
CREATE TABLE sales_data (
    sale_id NUMBER,
    product_name VARCHAR2(100),
    sale_amount NUMBER,
    sale_date DATE
);

-- Create sequence
CREATE SEQUENCE sales_seq START WITH 1;

-- Create trigger
CREATE OR REPLACE TRIGGER sales_audit_trg
AFTER DELETE ON sales_data
FOR EACH ROW
BEGIN
    INSERT INTO audit_log VALUES (:OLD.sale_id, 'DELETED', SYSDATE);
END;
/

-- Insert test data
INSERT INTO sales_data VALUES (sales_seq.NEXTVAL, 'Laptop', 1200, SYSDATE-10);
INSERT INTO sales_data VALUES (sales_seq.NEXTVAL, 'Mouse', 25, SYSDATE-5);
INSERT INTO sales_data VALUES (sales_seq.NEXTVAL, 'Keyboard', 75, SYSDATE-2);

-- Example 1: DELETE with condition
DELETE FROM sales_data WHERE sale_amount < 50;
-- Trigger fires, specific rows removed, can rollback

-- Example 2: TRUNCATE all data
TRUNCATE TABLE sales_data;
-- All data gone, structure remains, triggers don't fire, cannot rollback

-- Example 3: DROP entire table
DROP TABLE sales_data;
-- Everything gone including structure, triggers, indexes

-- ========================================
-- WHEN TO USE EACH COMMAND
-- ========================================

/*
USE DELETE WHEN:
- Need to remove specific rows based on conditions
- Need to remove ALL rows but maintain rollback capability
- Need ability to rollback the operation
- Working with small to medium datasets
- Need triggers to fire for audit/logging
- Need to maintain referential integrity
- Want to delete all data but slower than TRUNCATE (with rollback option)

USE TRUNCATE WHEN:
- Need to remove ALL rows from a table (faster than DELETE)
- Working with large datasets (performance critical)
- Don't need to rollback the operation  
- Don't need triggers to fire
- Want to reset table to empty state quickly
- Want to reset sequences/identity columns

USE DROP WHEN:
- Need to completely remove the table
- Table is no longer needed
- Want to free up all storage space
- Need to remove all associated objects (indexes, triggers, etc.)
- Restructuring database schema
*/

-- Clean up demonstration
-- DROP TABLE employees_demo; -- Already dropped above

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

Data state before query executes.

Output / Afterresult

After

COMMAND	CONDITION	ROWS_AFFECTED	ROLLBACK	SPEED
DELETE FROM employees WHERE dept_id=10	dept=10	2 rows	✅ Yes	Slow
DELETE FROM employees	no filter	4 rows	✅ Yes	Slow
TRUNCATE TABLE employees	N/A	4 rows	❌ No	Fast

TRUNCATE is fastest but irreversible — DELETE is flexible and rollbackable

Key Takeaway

DROP removes the entire table structure and data, DELETE removes specific rows with WHERE conditions and can be rolled back, TRUNCATE removes all rows quickly but cannot be rolled back.

Q49. How do you use ALTER TABLE to add, modify, or drop a column in Oracle?

DML & Data Manipulation

Q50. How do you delete all records from a table without deleting the table itself in Oracle?

Answer

Use TRUNCATE TABLE for fast deletion of all records, or DELETE FROM for conditional deletion with transaction control.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

CodeOracle SQL

SQL

-- Method 1: TRUNCATE TABLE (fastest, cannot be rolled back)
TRUNCATE TABLE employees_temp;

-- Method 2: DELETE FROM (slower, can be rolled back)
DELETE FROM employees_temp;

-- Method 3: DELETE with WHERE clause (delete specific records)
DELETE FROM employees_temp WHERE department_id = 10;

-- Method 4: Conditional deletion
DELETE FROM employees_temp 
WHERE hire_date < ADD_MONTHS(SYSDATE, -60);

-- Comparison: TRUNCATE vs DELETE
-- Create test table
CREATE TABLE test_employees AS SELECT * FROM employees;

-- Check record count
SELECT COUNT(*) FROM test_employees;

-- TRUNCATE characteristics:
-- 1. Faster than DELETE
-- 2. Resets high water mark
-- 3. Cannot be rolled back
-- 4. Resets SEQUENCE values
-- 5. Removes all rows without logging individual row deletions

TRUNCATE TABLE test_employees;

-- Recreate for DELETE example
INSERT INTO test_employees SELECT * FROM employees;

-- DELETE characteristics:
-- 1. Slower than TRUNCATE
-- 2. Can be rolled back
-- 3. Triggers fire for each row
-- 4. Generates undo information
-- 5. Can have WHERE clause

BEGIN
    DELETE FROM test_employees;
    -- Can rollback if needed
    ROLLBACK;
END;
/

-- Check count after rollback
SELECT COUNT(*) FROM test_employees;

-- Complete deletion with commit
DELETE FROM test_employees;
COMMIT;

-- Method 5: Using MERGE for conditional deletion
CREATE TABLE employees_to_delete (employee_id NUMBER);
INSERT INTO employees_to_delete VALUES (101);
INSERT INTO employees_to_delete VALUES (102);

MERGE INTO test_employees t
USING employees_to_delete d ON (t.employee_id = d.employee_id)
WHEN MATCHED THEN
    UPDATE SET first_name = NULL
    DELETE WHERE first_name IS NULL;

-- Method 6: Bulk delete with FORALL (PL/SQL)
DECLARE
    TYPE emp_id_array IS TABLE OF NUMBER;
    emp_ids emp_id_array;
BEGIN
    SELECT employee_id BULK COLLECT INTO emp_ids
    FROM employees
    WHERE department_id = 10;
    
    FORALL i IN emp_ids.FIRST..emp_ids.LAST
        DELETE FROM test_employees WHERE employee_id = emp_ids(i);
END;
/

-- Method 7: Delete with EXISTS subquery
DELETE FROM test_employees e1
WHERE EXISTS (
    SELECT 1 FROM departments d
    WHERE d.department_id = e1.department_id
    AND d.department_name = 'Sales'
);

-- Method 8: Delete with JOIN (using EXISTS)
DELETE FROM test_employees e
WHERE EXISTS (
    SELECT 1 FROM job_history jh
    WHERE jh.employee_id = e.employee_id
    AND jh.end_date IS NOT NULL
);

-- Method 9: Partitioned table truncation
-- TRUNCATE PARTITION (for partitioned tables)
-- ALTER TABLE sales_partitioned TRUNCATE PARTITION p_2023;

-- Method 10: Using ROWID for efficient deletion
DELETE FROM test_employees
WHERE ROWID IN (
    SELECT ROWID FROM test_employees
    WHERE department_id = 20
    AND ROWNUM <= 10
);

-- Performance monitoring
-- Check table statistics before and after
SELECT table_name, num_rows, blocks, empty_blocks
FROM user_tables
WHERE table_name = 'TEST_EMPLOYEES';

-- Update statistics after major deletions
BEGIN
    DBMS_STATS.GATHER_TABLE_STATS(USER, 'TEST_EMPLOYEES');
END;
/

-- Reclaim space after large deletions
-- ALTER TABLE test_employees SHRINK SPACE;

-- Clean up
DROP TABLE test_employees;
DROP TABLE employees_to_delete;

-- Best practices:
-- 1. Use TRUNCATE for complete table cleanup (faster)
-- 2. Use DELETE for conditional removal or when rollback needed
-- 3. Consider partitioning for large tables with regular cleanup needs
-- 4. Update statistics after major deletions
-- 5. Monitor space usage and consider shrinking tables when needed

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

Data state before query executes.

Output / Afterresult

After

EMP_ID	EMP_NAME	ACTION	RESULT
101	Alice Johnson	UPDATED (matched)	Salary → 55,000
102	Bob Smith	UPDATED (matched)	Salary → 68,200
999	New Hire	INSERTED (not matched)	New row added

MERGE: 2 rows updated + 1 inserted in one atomic statement

Key Takeaway

Use TRUNCATE TABLE for fast deletion of all records, or DELETE FROM for conditional deletion with transaction control.

Q51. What are the basic INSERT, UPDATE, and MERGE statements in Oracle?

Stored Procedures, Functions & Triggers

Q52. What is a stored procedure?

Q53. What is a trigger in SQL?

Q54. What is a cursor and when would you use it?

Q55. What are the benefits of using stored procedures?

Q56. Explain the use of triggers for maintaining database integrity.

Answer

Triggers play a crucial role in maintaining database integrity in SQL.

They are special types of stored procedures that automatically execute or 'fire' in response to certain events on a particular table or view in a database.

Triggers help in enforcing business rules and data integrity by automatically checking for certain conditions or changes in database data.

For example, a trigger prevents invalid data entry into a database by automatically reverting changes if the new data violates business rules.Triggers enhance the reliability of a database by maintaining a consistent state.

They ensure that all necessary changes occur as a result of data modifications.

This functionality becomes particularly valuable in complex databases where multiple interrelated actions need to occur in response to a single event.

Execute complex cascading actions seamlessly with triggers, provided the triggering event occurs.

This approach allows for the automation of routine tasks, thereby reducing the likelihood of human error and maintaining the overall integrity and reliability of the database.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

CodeOracle SQL

SQL

CREATE OR REPLACE TRIGGER trg_emp_before_insert
  BEFORE INSERT ON employees FOR EACH ROW
BEGIN
  IF :NEW.hire_date IS NULL THEN
    :NEW.hire_date := SYSDATE;
  END IF;
  IF :NEW.salary < 0 THEN
    RAISE_APPLICATION_ERROR(-20001,'Salary cannot be negative');
  END IF;
END;
/

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

Data state before query executes.

Output / Afterresult

After

EMP_ID	EMP_NAME	SALARY	HIRE_DATE
205	New Employee	50,000	20-APR-2026

Trigger auto-fired on INSERT — hire_date set to SYSDATE by BEFORE INSERT trigger

Key Takeaway

Triggers play a crucial role in maintaining database integrity in SQL.

Q57. How would you approach error handling in SQL stored procedures?

Answer

Approaching error handling in SQL stored procedures involves several key strategies.

Utilize the TRY...CATCH block to capture and manage errors.

This structure allows the execution of code in the TRY block, and if an error occurs, control is passed to the CATCH block where the error can be handled gracefully.

Implementing error logging is crucial, typically by inserting error details into a dedicated table.

This practice enables the tracking and analysis of errors over time.Another important aspect is the use of RAISERROR or THROW statements to generate custom error messages, allowing for more specific and informative feedback about issues encountered in the stored procedure.

Ensure proper transaction management by using COMMIT and ROLLBACK within the TRY...CATCH block.

This ensures that the database remains consistent and any changes made during the procedure are either fully committed or rolled back, depending on the success or failure of the procedure.

Employing these methods ensures robust error handling in SQL stored procedures, enhancing reliability and maintainability of the database application.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

CodeOracle SQL

SQL

CREATE OR REPLACE PROCEDURE prc_give_raise(
  p_dept_id IN NUMBER, p_pct IN NUMBER, p_rows OUT NUMBER
) IS
BEGIN
  UPDATE employees SET salary = salary * (1 + p_pct/100)
  WHERE  department_id = p_dept_id;
  p_rows := SQL%ROWCOUNT; COMMIT;
END;
/

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

Data state before query executes.

Output / Afterresult

After

EMP_ID	EMP_NAME	OLD_SALARY	NEW_SALARY
101	Alice Johnson	50,000	55,000
103	Carol White	55,000	60,500

2 rows updated by stored procedure — 10% raise applied to dept 10

Key Takeaway

Approaching error handling in SQL stored procedures involves several key strategies.

Q58. What is a stored procedure in Oracle PL/SQL and how do you create one?

Answer

A stored procedure is a precompiled group of SQL and PL/SQL statements stored in the database.

It can be executed multiple times and can accept parameters.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

CodeOracle SQL

SQL

-- Create a simple stored procedure
CREATE OR REPLACE PROCEDURE update_employee_salary(
    p_employee_id IN NUMBER,
    p_percentage IN NUMBER
) IS
    v_current_salary NUMBER;
    v_new_salary NUMBER;
BEGIN
    -- Get current salary
    SELECT salary INTO v_current_salary
    FROM employees
    WHERE employee_id = p_employee_id;
    
    -- Calculate new salary
    v_new_salary := v_current_salary * (1 + p_percentage/100);
    
    -- Update salary
    UPDATE employees
    SET salary = v_new_salary
    WHERE employee_id = p_employee_id;
    
    COMMIT;
    
    DBMS_OUTPUT.PUT_LINE('Salary updated from ' || v_current_salary || ' to ' || v_new_salary);
    
EXCEPTION
    WHEN NO_DATA_FOUND THEN
        DBMS_OUTPUT.PUT_LINE('Employee not found');
    WHEN OTHERS THEN
        ROLLBACK;
        DBMS_OUTPUT.PUT_LINE('Error: ' || SQLERRM);
END;
/

-- Procedure with OUT parameter
CREATE OR REPLACE PROCEDURE get_employee_details(
    p_employee_id IN NUMBER,
    p_name OUT VARCHAR2,
    p_salary OUT NUMBER,
    p_department OUT VARCHAR2
) IS
BEGIN
    SELECT e.first_name || ' ' || e.last_name,
           e.salary,
           d.department_name
    INTO p_name, p_salary, p_department
    FROM employees e
    JOIN departments d ON e.department_id = d.department_id
    WHERE e.employee_id = p_employee_id;
EXCEPTION
    WHEN NO_DATA_FOUND THEN
        p_name := 'Not Found';
        p_salary := 0;
        p_department := 'N/A';
END;
/

-- Execute stored procedures
BEGIN
    update_employee_salary(101, 10); -- Give 10% raise to employee 101
END;
/

-- Execute procedure with OUT parameters
DECLARE
    v_name VARCHAR2(100);
    v_salary NUMBER;
    v_dept VARCHAR2(100);
BEGIN
    get_employee_details(101, v_name, v_salary, v_dept);
    DBMS_OUTPUT.PUT_LINE('Employee: ' || v_name);
    DBMS_OUTPUT.PUT_LINE('Salary: ' || v_salary);
    DBMS_OUTPUT.PUT_LINE('Department: ' || v_dept);
END;
/

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

Data state before query executes.

Output / Afterresult

After

EXCEPTION	WHEN_RAISED	HANDLED_BY	OUTCOME
NO_DATA_FOUND	SELECT INTO = 0 rows	WHEN NO_DATA_FOUND	Continue after message
TOO_MANY_ROWS	SELECT INTO > 1 row	WHEN OTHERS	Rollback + log

Exceptions prevent unhandled errors from crashing the PL/SQL block

Key Takeaway

A stored procedure is a precompiled group of SQL and PL/SQL statements stored in the database.

Q59. What is a trigger in Oracle and how do you create one?

Answer

A trigger is a special type of PL/SQL block that automatically executes (fires) in response to specific database events such as INSERT, UPDATE, or DELETE operations.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

CodeOracle SQL

SQL

-- 1. BEFORE INSERT TRIGGER
CREATE OR REPLACE TRIGGER trg_emp_before_insert
BEFORE INSERT ON employees
FOR EACH ROW
BEGIN
    -- Auto-generate employee ID if not provided
    IF :NEW.employee_id IS NULL THEN
        SELECT emp_seq.NEXTVAL INTO :NEW.employee_id FROM dual;
    END IF;
    
    -- Set hire date to current date if not provided
    IF :NEW.hire_date IS NULL THEN
        :NEW.hire_date := SYSDATE;
    END IF;
    
    -- Convert email to uppercase
    :NEW.email := UPPER(:NEW.email);
END;
/

-- 2. AFTER UPDATE TRIGGER
CREATE OR REPLACE TRIGGER trg_emp_after_update
AFTER UPDATE ON employees
FOR EACH ROW
BEGIN
    -- Log salary changes
    IF :OLD.salary != :NEW.salary THEN
        INSERT INTO salary_audit_log (
            employee_id, 
            old_salary, 
            new_salary, 
            change_date, 
            changed_by
        ) VALUES (
            :NEW.employee_id,
            :OLD.salary,
            :NEW.salary,
            SYSDATE,
            USER
        );
    END IF;
END;
/

-- 3. BEFORE DELETE TRIGGER
CREATE OR REPLACE TRIGGER trg_emp_before_delete
BEFORE DELETE ON employees
FOR EACH ROW
BEGIN
    -- Prevent deletion of employees with high salary
    IF :OLD.salary > 100000 THEN
        RAISE_APPLICATION_ERROR(-20001, 
            'Cannot delete high-salary employee: ' || :OLD.first_name || ' ' || :OLD.last_name);
    END IF;
    
    -- Archive employee data before deletion
    INSERT INTO employees_archive VALUES :OLD;
END;
/

-- 4. STATEMENT-LEVEL TRIGGER
CREATE OR REPLACE TRIGGER trg_emp_statement_audit
AFTER INSERT OR UPDATE OR DELETE ON employees
BEGIN
    INSERT INTO table_audit_log (
        table_name,
        operation,
        timestamp,
        user_name
    ) VALUES (
        'EMPLOYEES',
        CASE 
            WHEN INSERTING THEN 'INSERT'
            WHEN UPDATING THEN 'UPDATE'
            WHEN DELETING THEN 'DELETE'
        END,
        SYSDATE,
        USER
    );
END;
/

-- 5. COMPOUND TRIGGER (Oracle 11g+)
CREATE OR REPLACE TRIGGER trg_emp_compound
FOR INSERT OR UPDATE OR DELETE ON employees
COMPOUND TRIGGER
    
    -- Declaration section
    TYPE emp_id_list_t IS TABLE OF employees.employee_id%TYPE;
    l_emp_ids emp_id_list_t := emp_id_list_t();
    
    -- BEFORE STATEMENT
    BEFORE STATEMENT IS
    BEGIN
        DBMS_OUTPUT.PUT_LINE('Starting DML operation on employees table');
    END BEFORE STATEMENT;
    
    -- BEFORE EACH ROW
    BEFORE EACH ROW IS
    BEGIN
        IF INSERTING THEN
            :NEW.created_date := SYSDATE;
        ELSIF UPDATING THEN
            :NEW.modified_date := SYSDATE;
        END IF;
    END BEFORE EACH ROW;
    
    -- AFTER EACH ROW
    AFTER EACH ROW IS
    BEGIN
        l_emp_ids.EXTEND;
        l_emp_ids(l_emp_ids.COUNT) := :NEW.employee_id;
    END AFTER EACH ROW;
    
    -- AFTER STATEMENT
    AFTER STATEMENT IS
    BEGIN
        DBMS_OUTPUT.PUT_LINE('Processed ' || l_emp_ids.COUNT || ' employees');
        -- Bulk processing of collected employee IDs
        FOR i IN 1..l_emp_ids.COUNT LOOP
            -- Update related tables or perform batch operations
            NULL;
        END LOOP;
    END AFTER STATEMENT;
    
END trg_emp_compound;
/

-- 6. DDL TRIGGER
CREATE OR REPLACE TRIGGER trg_ddl_audit
AFTER DDL ON SCHEMA
BEGIN
    INSERT INTO ddl_audit_log (
        username,
        ddl_date,
        ddl_type,
        object_type,
        object_name,
        sql_text
    ) VALUES (
        USER,
        SYSDATE,
        SYS.DICTIONARY_OBJ_TYPE,
        SYS.DICTIONARY_OBJ_OWNER,
        SYS.DICTIONARY_OBJ_NAME,
        SYS.LOGIN_USER
    );
END;
/

-- Trigger Management Commands
-- Enable/Disable trigger
ALTER TRIGGER trg_emp_before_insert DISABLE;
ALTER TRIGGER trg_emp_before_insert ENABLE;

-- Drop trigger
DROP TRIGGER trg_emp_before_insert;

-- View trigger information
SELECT trigger_name, table_name, triggering_event, status
FROM user_triggers
WHERE table_name = 'EMPLOYEES';

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

Data state before query executes.

Output / Afterresult

After

EMP_ID	EMP_NAME	DEPT_ID	SALARY
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	20	70,000

All 4 rows returned

Key Takeaway

A trigger is a special type of PL/SQL block that automatically executes (fires) in response to specific database events such as INSERT, UPDATE, or DELETE operations.

Q60. How do you implement database triggers for audit trails in Oracle?

Answer

Create triggers that automatically capture DML operations (INSERT, UPDATE, DELETE) and store audit information in separate audit tables.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

CodeOracle SQL

SQL

-- Create audit table
CREATE TABLE employee_audit (
   audit_id NUMBER GENERATED ALWAYS AS IDENTITY,
   table_name VARCHAR2(30),
   operation VARCHAR2(10),
   employee_id NUMBER,
   old_values CLOB,
   new_values CLOB,
   changed_by VARCHAR2(100),
   change_date TIMESTAMP DEFAULT SYSTIMESTAMP,
   session_id NUMBER,
   terminal VARCHAR2(100),
   program VARCHAR2(100)
);

-- Create audit trigger for employees table
CREATE OR REPLACE TRIGGER trg_employee_audit
AFTER INSERT OR UPDATE OR DELETE ON employees
FOR EACH ROW
DECLARE
   v_operation VARCHAR2(10);
   v_old_values CLOB;
   v_new_values CLOB;
BEGIN
   -- Determine operation type
   IF INSERTING THEN
       v_operation := 'INSERT';
       v_old_values := NULL;
       v_new_values := 'employee_id:' || :NEW.employee_id || 
                      ',first_name:' || :NEW.first_name ||
                      ',last_name:' || :NEW.last_name ||
                      ',email:' || :NEW.email ||
                      ',salary:' || :NEW.salary ||
                      ',department_id:' || :NEW.department_id;
   ELSIF UPDATING THEN
       v_operation := 'UPDATE';
       v_old_values := 'employee_id:' || :OLD.employee_id || 
                      ',first_name:' || :OLD.first_name ||
                      ',last_name:' || :OLD.last_name ||
                      ',email:' || :OLD.email ||
                      ',salary:' || :OLD.salary ||
                      ',department_id:' || :OLD.department_id;
       v_new_values := 'employee_id:' || :NEW.employee_id || 
                      ',first_name:' || :NEW.first_name ||
                      ',last_name:' || :NEW.last_name ||
                      ',email:' || :NEW.email ||
                      ',salary:' || :NEW.salary ||
                      ',department_id:' || :NEW.department_id;
   ELSIF DELETING THEN
       v_operation := 'DELETE';
       v_old_values := 'employee_id:' || :OLD.employee_id || 
                      ',first_name:' || :OLD.first_name ||
                      ',last_name:' || :OLD.last_name ||
                      ',email:' || :OLD.email ||
                      ',salary:' || :OLD.salary ||
                      ',department_id:' || :OLD.department_id;
       v_new_values := NULL;
   END IF;
   
   -- Insert audit record
   INSERT INTO employee_audit (
       table_name, operation, employee_id, old_values, new_values,
       changed_by, session_id, terminal, program
   ) VALUES (
       'EMPLOYEES', v_operation, 
       COALESCE(:NEW.employee_id, :OLD.employee_id),
       v_old_values, v_new_values,
       USER, SYS_CONTEXT('USERENV', 'SESSIONID'),
       SYS_CONTEXT('USERENV', 'TERMINAL'),
       SYS_CONTEXT('USERENV', 'MODULE')
   );
END;
/

-- Advanced audit trigger with JSON format
CREATE OR REPLACE TRIGGER trg_employee_audit_json
AFTER INSERT OR UPDATE OR DELETE ON employees
FOR EACH ROW
DECLARE
   v_operation VARCHAR2(10);
   v_old_json CLOB;
   v_new_json CLOB;
BEGIN
   IF INSERTING THEN
       v_operation := 'INSERT';
       v_new_json := JSON_OBJECT(
           'employee_id' VALUE :NEW.employee_id,
           'first_name' VALUE :NEW.first_name,
           'last_name' VALUE :NEW.last_name,
           'email' VALUE :NEW.email,
           'salary' VALUE :NEW.salary,
           'department_id' VALUE :NEW.department_id,
           'hire_date' VALUE TO_CHAR(:NEW.hire_date, 'YYYY-MM-DD'),
           'manager_id' VALUE :NEW.manager_id
       );
   ELSIF UPDATING THEN
       v_operation := 'UPDATE';
       v_old_json := JSON_OBJECT(
           'employee_id' VALUE :OLD.employee_id,
           'first_name' VALUE :OLD.first_name,
           'last_name' VALUE :OLD.last_name,
           'email' VALUE :OLD.email,
           'salary' VALUE :OLD.salary,
           'department_id' VALUE :OLD.department_id,
           'hire_date' VALUE TO_CHAR(:OLD.hire_date, 'YYYY-MM-DD'),
           'manager_id' VALUE :OLD.manager_id
       );
       v_new_json := JSON_OBJECT(
           'employee_id' VALUE :NEW.employee_id,
           'first_name' VALUE :NEW.first_name,
           'last_name' VALUE :NEW.last_name,
           'email' VALUE :NEW.email,
           'salary' VALUE :NEW.salary,
           'department_id' VALUE :NEW.department_id,
           'hire_date' VALUE TO_CHAR(:NEW.hire_date, 'YYYY-MM-DD'),
           'manager_id' VALUE :NEW.manager_id
       );
   ELSIF DELETING THEN
       v_operation := 'DELETE';
       v_old_json := JSON_OBJECT(
           'employee_id' VALUE :OLD.employee_id,
           'first_name' VALUE :OLD.first_name,
           'last_name' VALUE :OLD.last_name,
           'email' VALUE :OLD.email,
           'salary' VALUE :OLD.salary,
           'department_id' VALUE :OLD.department_id,
           'hire_date' VALUE TO_CHAR(:OLD.hire_date, 'YYYY-MM-DD'),
           'manager_id' VALUE :OLD.manager_id
       );
   END IF;
   
   INSERT INTO employee_audit (
       table_name, operation, employee_id, old_values, new_values,
       changed_by, session_id, terminal, program
   ) VALUES (
       'EMPLOYEES', v_operation, 
       COALESCE(:NEW.employee_id, :OLD.employee_id),
       v_old_json, v_new_json,
       USER, SYS_CONTEXT('USERENV', 'SESSIONID'),
       SYS_CONTEXT('USERENV', 'TERMINAL'),
       SYS_CONTEXT('USERENV', 'MODULE')
   );
END;
/

-- Selective column audit trigger (only track specific changes)
CREATE OR REPLACE TRIGGER trg_employee_salary_audit
AFTER UPDATE OF salary, department_id ON employees
FOR EACH ROW
WHEN (OLD.salary != NEW.salary OR OLD.department_id != NEW.department_id)
BEGIN
   INSERT INTO employee_audit (
       table_name, operation, employee_id, old_values, new_values, changed_by
   ) VALUES (
       'EMPLOYEES', 'UPDATE', :NEW.employee_id,
       'salary:' || :OLD.salary || ',department_id:' || :OLD.department_id,
       'salary:' || :NEW.salary || ',department_id:' || :NEW.department_id,
       USER
   );
END;
/

-- Query audit trail
SELECT audit_id, operation, employee_id, changed_by, change_date,
      old_values, new_values
FROM employee_audit
WHERE employee_id = 101
ORDER BY change_date DESC;

-- Audit trail analysis queries
-- Most active users
SELECT changed_by, COUNT(*) as change_count,
      COUNT(CASE WHEN operation = 'INSERT' THEN 1 END) as inserts,
      COUNT(CASE WHEN operation = 'UPDATE' THEN 1 END) as updates,
      COUNT(CASE WHEN operation = 'DELETE' THEN 1 END) as deletes
FROM employee_audit
WHERE change_date >= TRUNC(SYSDATE) - 30
GROUP BY changed_by
ORDER BY change_count DESC;

-- Changes by time period
SELECT TRUNC(change_date) as change_day,
      COUNT(*) as total_changes,
      COUNT(DISTINCT employee_id) as employees_affected
FROM employee_audit
WHERE change_date >= TRUNC(SYSDATE) - 7
GROUP BY TRUNC(change_date)
ORDER BY change_day;

-- Salary change history for specific employee
SELECT change_date, changed_by,
      JSON_VALUE(old_values, '$.salary') as old_salary,
      JSON_VALUE(new_values, '$.salary') as new_salary,
      JSON_VALUE(new_values, '$.salary') - JSON_VALUE(old_values, '$.salary') as salary_increase
FROM employee_audit
WHERE employee_id = 101
 AND operation = 'UPDATE'
 AND JSON_EXISTS(old_values, '$.salary')
 AND JSON_VALUE(old_values, '$.salary') != JSON_VALUE(new_values, '$.salary')
ORDER BY change_date;

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

Data state before query executes.

Output / Afterresult

After

EMPLOYEE	MANAGER
Bob Smith	Alice Johnson
Carol White	Alice Johnson
Dave Brown	Bob Smith

3 rows — self join pairs each employee with their manager's name

Key Takeaway

Create triggers that automatically capture DML operations (INSERT, UPDATE, DELETE) and store audit information in separate audit tables.

Q61. How do you implement custom aggregate functions in Oracle?

Answer

Create custom aggregate functions using PL/SQL object types with ODCIAggregate interface methods for complex aggregation logic.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

CodeOracle SQL

SQL

-- Create object type for custom aggregate
CREATE OR REPLACE TYPE weighted_avg_type AS OBJECT (
   total_weighted_value NUMBER,
   total_weight NUMBER,
   
   -- Required methods for aggregate interface
   STATIC FUNCTION ODCIAggregateInitialize(
       ctx IN OUT weighted_avg_type
   ) RETURN NUMBER,
   
   MEMBER FUNCTION ODCIAggregateIterate(
       self IN OUT weighted_avg_type,
       value IN NUMBER,
       weight IN NUMBER
   ) RETURN NUMBER,
   
   MEMBER FUNCTION ODCIAggregateMerge(
       self IN OUT weighted_avg_type,
       ctx IN weighted_avg_type
   ) RETURN NUMBER,
   
   MEMBER FUNCTION ODCIAggregateTerminate(
       self IN weighted_avg_type,
       returnValue OUT NUMBER,
       flags IN NUMBER
   ) RETURN NUMBER
);
/

-- Implement the object type body
CREATE OR REPLACE TYPE BODY weighted_avg_type IS
   
   STATIC FUNCTION ODCIAggregateInitialize(
       ctx IN OUT weighted_avg_type
   ) RETURN NUMBER IS
   BEGIN
       ctx := weighted_avg_type(0, 0);
       RETURN ODCIConst.Success;
   END;
   
   MEMBER FUNCTION ODCIAggregateIterate(
       self IN OUT weighted_avg_type,
       value IN NUMBER,
       weight IN NUMBER
   ) RETURN NUMBER IS
   BEGIN
       IF value IS NOT NULL AND weight IS NOT NULL THEN
           self.total_weighted_value := self.total_weighted_value + (value * weight);
           self.total_weight := self.total_weight + weight;
       END IF;
       RETURN ODCIConst.Success;
   END;
   
   MEMBER FUNCTION ODCIAggregateMerge(
       self IN OUT weighted_avg_type,
       ctx IN weighted_avg_type
   ) RETURN NUMBER IS
   BEGIN
       self.total_weighted_value := self.total_weighted_value + ctx.total_weighted_value;
       self.total_weight := self.total_weight + ctx.total_weight;
       RETURN ODCIConst.Success;
   END;
   
   MEMBER FUNCTION ODCIAggregateTerminate(
       self IN weighted_avg_type,
       returnValue OUT NUMBER,
       flags IN NUMBER
   ) RETURN NUMBER IS
   BEGIN
       IF self.total_weight > 0 THEN
           returnValue := self.total_weighted_value / self.total_weight;
       ELSE
           returnValue := NULL;
       END IF;
       RETURN ODCIConst.Success;
   END;
   
END;
/

-- Create the aggregate function
CREATE OR REPLACE FUNCTION weighted_average(
   value NUMBER,
   weight NUMBER
) RETURN NUMBER
AGGREGATE USING weighted_avg_type;
/

-- Test the custom aggregate function
SELECT department_id,
      AVG(salary) as simple_average,
      weighted_average(salary, years_of_service) as weighted_avg_salary
FROM (
   SELECT department_id, salary,
          EXTRACT(YEAR FROM SYSDATE) - EXTRACT(YEAR FROM hire_date) as years_of_service
   FROM employees
)
GROUP BY department_id;

-- Advanced custom aggregate: String concatenation with separator
CREATE OR REPLACE TYPE concat_type AS OBJECT (
   result_string CLOB,
   separator VARCHAR2(10),
   
   STATIC FUNCTION ODCIAggregateInitialize(
       ctx IN OUT concat_type
   ) RETURN NUMBER,
   
   MEMBER FUNCTION ODCIAggregateIterate(
       self IN OUT concat_type,
       input_string IN VARCHAR2,
       sep IN VARCHAR2 DEFAULT ','
   ) RETURN NUMBER,
   
   MEMBER FUNCTION ODCIAggregateMerge(
       self IN OUT concat_type,
       ctx IN concat_type
   ) RETURN NUMBER,
   
   MEMBER FUNCTION ODCIAggregateTerminate(
       self IN concat_type,
       returnValue OUT CLOB,
       flags IN NUMBER
   ) RETURN NUMBER
);
/

CREATE OR REPLACE TYPE BODY concat_type IS
   
   STATIC FUNCTION ODCIAggregateInitialize(
       ctx IN OUT concat_type
   ) RETURN NUMBER IS
   BEGIN
       ctx := concat_type(EMPTY_CLOB(), ',');
       DBMS_LOB.CREATETEMPORARY(ctx.result_string, TRUE);
       RETURN ODCIConst.Success;
   END;
   
   MEMBER FUNCTION ODCIAggregateIterate(
       self IN OUT concat_type,
       input_string IN VARCHAR2,
       sep IN VARCHAR2 DEFAULT ','
   ) RETURN NUMBER IS
   BEGIN
       IF input_string IS NOT NULL THEN
           IF DBMS_LOB.GETLENGTH(self.result_string) > 0 THEN
               DBMS_LOB.APPEND(self.result_string, NVL(sep, self.separator));
           END IF;
           DBMS_LOB.APPEND(self.result_string, input_string);
           self.separator := NVL(sep, self.separator);
       END IF;
       RETURN ODCIConst.Success;
   END;
   
   MEMBER FUNCTION ODCIAggregateMerge(
       self IN OUT concat_type,
       ctx IN concat_type
   ) RETURN NUMBER IS
   BEGIN
       IF DBMS_LOB.GETLENGTH(ctx.result_string) > 0 THEN
           IF DBMS_LOB.GETLENGTH(self.result_string) > 0 THEN
               DBMS_LOB.APPEND(self.result_string, self.separator);
           END IF;
           DBMS_LOB.APPEND(self.result_string, ctx.result_string);
       END IF;
       RETURN ODCIConst.Success;
   END;
   
   MEMBER FUNCTION ODCIAggregateTerminate(
       self IN concat_type,
       returnValue OUT CLOB,
       flags IN NUMBER
   ) RETURN NUMBER IS
   BEGIN
       returnValue := self.result_string;
       RETURN ODCIConst.Success;
   END;
   
END;
/

-- Create custom concatenate function
CREATE OR REPLACE FUNCTION custom_concat(
   input_string VARCHAR2,
   separator VARCHAR2 DEFAULT ','
) RETURN CLOB
AGGREGATE USING concat_type;
/

-- Test custom concatenation
SELECT department_id,
      custom_concat(first_name || ' ' || last_name, '; ') as employee_list
FROM employees
GROUP BY department_id;

-- Geometric mean aggregate function
CREATE OR REPLACE TYPE geom_mean_type AS OBJECT (
   product_log NUMBER,
   count_values NUMBER,
   
   STATIC FUNCTION ODCIAggregateInitialize(
       ctx IN OUT geom_mean_type
   ) RETURN NUMBER,
   
   MEMBER FUNCTION ODCIAggregateIterate(
       self IN OUT geom_mean_type,
       value IN NUMBER
   ) RETURN NUMBER,
   
   MEMBER FUNCTION ODCIAggregateMerge(
       self IN OUT geom_mean_type,
       ctx IN geom_mean_type
   ) RETURN NUMBER,
   
   MEMBER FUNCTION ODCIAggregateTerminate(
       self IN geom_mean_type,
       returnValue OUT NUMBER,
       flags IN NUMBER
   ) RETURN NUMBER
);
/

CREATE OR REPLACE TYPE BODY geom_mean_type IS
   
   STATIC FUNCTION ODCIAggregateInitialize(
       ctx IN OUT geom_mean_type
   ) RETURN NUMBER IS
   BEGIN
       ctx := geom_mean_type(0, 0);
       RETURN ODCIConst.Success;
   END;
   
   MEMBER FUNCTION ODCIAggregateIterate(
       self IN OUT geom_mean_type,
       value IN NUMBER
   ) RETURN NUMBER IS
   BEGIN
       IF value IS NOT NULL AND value > 0 THEN
           self.product_log := self.product_log + LN(value);
           self.count_values := self.count_values + 1;
       END IF;
       RETURN ODCIConst.Success;
   END;
   
   MEMBER FUNCTION ODCIAggregateMerge(
       self IN OUT geom_mean_type,
       ctx IN geom_mean_type
   ) RETURN NUMBER IS
   BEGIN
       self.product_log := self.product_log + ctx.product_log;
       self.count_values := self.count_values + ctx.count_values;
       RETURN ODCIConst.Success;
   END;
   
   MEMBER FUNCTION ODCIAggregateTerminate(
       self IN geom_mean_type,
       returnValue OUT NUMBER,
       flags IN NUMBER
   ) RETURN NUMBER IS
   BEGIN
       IF self.count_values > 0 THEN
           returnValue := EXP(self.product_log / self.count_values);
       ELSE
           returnValue := NULL;
       END IF;
       RETURN ODCIConst.Success;
   END;
   
END;
/

CREATE OR REPLACE FUNCTION geometric_mean(value NUMBER) 
RETURN NUMBER AGGREGATE USING geom_mean_type;
/

-- Test geometric mean
SELECT department_id,
      ROUND(AVG(salary), 2) as arithmetic_mean,
      ROUND(geometric_mean(salary), 2) as geometric_mean_salary
FROM employees
WHERE salary > 0
GROUP BY department_id;

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

Data state before query executes.

Output / Afterresult

After

DEPT_ID	EMP_COUNT	AVG_SALARY	MAX_SALARY
10	2	52,500	55,000
20	2	66,000	70,000

2 groups — Alice+Carol in dept 10, Bob+Dave in dept 20

Key Takeaway

Create custom aggregate functions using PL/SQL object types with ODCIAggregate interface methods for complex aggregation logic.

Q62. What is the difference between a function and a procedure in Oracle PL/SQL?

Q63. What are the different types of triggers in Oracle, based on timing and level?

Transactions, Concurrency & Locking

Q64. What is a transaction in SQL?

Q65. What are ACID properties in a database?

Answer

ACID properties in a database are fundamental principles that ensure the reliability and consistency of data in SQL systems.

These properties, which stand for Atomicity, Consistency, Isolation, and Durability, are crucial for maintaining data integrity.Atomicity guarantees that database transactions are treated as a single unit, and they either fully succeed or fail.

This means that all the operations within a transaction are executed entirely, or none of them are.Consistency ensures that a database remains in a valid state before and after a transaction.

It means that data changes made by a transaction must adhere to predefined rules and constraints.Isolation prevents concurrent transactions from interfering with each other.

SQL databases use locking mechanisms to isolate transactions and maintain their independence.Durability guarantees that once a transaction is committed, its changes are permanent and will survive any system failures.

In SQL, this means that the data remains intact even in the face of power outages or crashes.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

CodeOracle SQL

SQL

SELECT * FROM employees WHERE ROWNUM <= 5;

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

Data state before query executes.

Output / Afterresult

After

PROPERTY	GUARANTEED_BY	EXAMPLE
Atomicity	Transaction rollback	Transfer: both debit+credit or neither
Consistency	Constraints + triggers	Balance never goes negative
Isolation	Locking / MVCC	Session 2 sees no partial changes
Durability	Redo logs	Committed data survives power failure

ACID properties are the foundation of reliable relational database transactions

Key Takeaway

ACID properties in a database are fundamental principles that ensure the reliability and consistency of data in SQL systems.

Q66. Explain the concept of ACID compliance in the context of distributed databases.

Answer

ACID compliance, in the context of distributed databases, ensures that database transactions are executed reliably and consistently.

ACID stands for Atomicity, Consistency, Isolation, and Durability.Atomicity guarantees that a transaction is treated as a single, indivisible unit.

All its operations are either completed or aborted as a whole, ensuring data integrity.

Consistency ensures that a transaction takes the database from one consistent state to another.

It maintains data validity and integrity throughout the process.Isolation guarantees that multiple concurrent transactions do not interfere with each other.

Each transaction is executed in isolation, as if it were the only one running.

Durability ensures that once a transaction is committed, its changes are permanent and survive system failures.ACID compliance in distributed databases ensures data reliability, consistency, and integrity, even in a distributed and potentially unreliable environment.

It's crucial for maintaining data quality and correctness in SQL-based distributed systems.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

CodeOracle SQL

SQL

SELECT * FROM employees WHERE ROWNUM <= 5;

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

Data state before query executes.

Output / Afterresult

After

PROPERTY	GUARANTEED_BY	EXAMPLE
Atomicity	Transaction rollback	Transfer: both debit+credit or neither
Consistency	Constraints + triggers	Balance never goes negative
Isolation	Locking / MVCC	Session 2 sees no partial changes
Durability	Redo logs	Committed data survives power failure

ACID properties are the foundation of reliable relational database transactions

Key Takeaway

ACID compliance, in the context of distributed databases, ensures that database transactions are executed reliably and consistently.

Q67. How do you handle deadlock situations in SQL?

Answer

Handling deadlock situations in SQL involves detecting and resolving conflicts that occur when multiple transactions are competing for the same resources simultaneously.SQL provides mechanisms to handle deadlock situations efficiently.

One approach is to use locks, such as shared locks and exclusive locks, to control access to data.

A transaction acquires a lock, when it requests access to a resource, preventing other transactions from accessing the same resource simultaneously.

A deadlock occurs, If two or more transactions are blocked waiting for each other's resources.To address deadlock situations one common method is to implement a timeout mechanism, where transactions are set to automatically release their locks after a predefined period.

This helps prevent indefinite blocking.

SQL supports deadlock detection algorithms that periodically check for deadlock conditions.

The system automatically terminates one of the conflicting transactions to resolve the issue, If a deadlock is detected.SQL allows for the definition of deadlock priorities, where certain transactions are given higher precedence in case of conflicts, ensuring that critical processes continue execution while less critical ones may be rolled back.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

CodeOracle SQL

SQL

SELECT * FROM employees WHERE ROWNUM <= 5;

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

Data state before query executes.

Output / Afterresult

After

SESSION	HOLDS_LOCK	WAITING_FOR	ORACLE_ACTION
Session 1	Acc 1001	Acc 1002	✅ Allowed to commit
Session 2	Acc 1002	Acc 1001	❌ ORA-00060: deadlock — rolled back

Oracle detects the cycle automatically and rolls back the victim session

Key Takeaway

Handling deadlock situations in SQL involves detecting and resolving conflicts that occur when multiple transactions are competing for the same resources simultaneously.

Q68. How do you manage transactions in a distributed database system?

Q69. How does SQL handle data integrity and consistency?

Q70. How does SQL handle concurrency and what are the challenges involved?

Answer

SQL handles concurrency through a set of mechanisms that ensure data integrity and consistency when multiple transactions occur simultaneously.

The primary challenge in managing concurrency in SQL is to balance the need for simultaneous access to data with the necessity of maintaining data accuracy and consistency.

Transactions, which are sequences of operations performed as a single logical unit, play a crucial role in this balance.

SQL databases use locking and isolation levels as key tools to manage concurrent access.

Locking prevents multiple transactions from accessing the same data concurrently, thus avoiding conflicts and data corruption.

However, excessive locking can lead to performance bottlenecks, as it restricts access to data.Isolation levels in SQL determine how much a transaction is isolated from other transactions.

Higher isolation levels provide greater data integrity but at the cost of reduced concurrency and potential performance issues like deadlocks, where two or more transactions are waiting indefinitely for one another to release locks.

SQL databases offer different isolation levels such as Read Uncommitted, Read Committed, Repeatable Read, and Serializable, each providing a different balance between data integrity and concurrency.

It is essential to choose the appropriate isolation level, considering the specific requirements of the application.

This choice impacts how transactions interact, for optimal concurrency management, ensuring that data remains consistent and reliable even when accessed by multiple users or processes simultaneously.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

CodeOracle SQL

SQL

SELECT * FROM employees WHERE ROWNUM <= 5;

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

Data state before query executes.

Output / Afterresult

After

EMP_ID	EMP_NAME	SALARY
101	Alice Johnson	50,000
102	Bob Smith	62,000
103	Carol White	55,000

Top 3 rows — ROWNUM/FETCH FIRST limits result set

Key Takeaway

SQL handles concurrency through a set of mechanisms that ensure data integrity and consistency when multiple transactions occur simultaneously.

Q71. Explain the ACID properties in Oracle Database.

Answer

ACID stands for Atomicity, Consistency, Isolation, and Durability.

These properties ensure that database transactions are processed reliably in Oracle.

Sample Dataused in example

accounts

account_id	owner	balance
1001	Alice	5,000
1002	Bob	3,200
1003	Carol	8,750

CodeOracle SQL

SQL

-- ACID Properties Demonstration in Oracle

-- 1. ATOMICITY - All or nothing principle
BEGIN
    -- Transfer money between accounts (atomic transaction)
    UPDATE accounts SET balance = balance - 1000 WHERE account_id = 'ACC001';
    UPDATE accounts SET balance = balance + 1000 WHERE account_id = 'ACC002';
    
    -- If any statement fails, entire transaction is rolled back
    COMMIT; -- All changes saved together
EXCEPTION
    WHEN OTHERS THEN
        ROLLBACK; -- All changes undone together
        RAISE;
END;
/

-- 2. CONSISTENCY - Database remains in valid state
CREATE OR REPLACE TRIGGER trg_account_balance_check
BEFORE UPDATE ON accounts
FOR EACH ROW
BEGIN
    -- Ensure balance never goes negative (business rule)
    IF :NEW.balance < 0 THEN
        RAISE_APPLICATION_ERROR(-20001, 'Balance cannot be negative');
    END IF;
END;
/

-- 3. ISOLATION - Transactions don't interfere with each other
-- Session 1:
BEGIN
    UPDATE employees SET salary = salary * 1.1 WHERE department_id = 10;
    -- Transaction not yet committed, changes not visible to other sessions
    COMMIT;
END;
/

-- Different isolation levels in Oracle
-- READ COMMITTED (default)
ALTER SESSION SET ISOLATION_LEVEL = READ COMMITTED;

-- SERIALIZABLE
ALTER SESSION SET ISOLATION_LEVEL = SERIALIZABLE;

-- Demonstrate isolation with locks
SELECT object_name, object_type, session_id, oracle_username, locked_mode
FROM v$locked_object lo
JOIN dba_objects do ON lo.object_id = do.object_id;

-- 4. DURABILITY - Committed changes survive system failures
-- Oracle ensures durability through:
-- - Redo logs
-- - Database checkpoints
-- - Archive logs

-- Check redo log status
SELECT group#, status, archived, bytes/1024/1024 as size_mb
FROM v$log;

-- Force log switch for durability
ALTER SYSTEM SWITCH LOGFILE;

-- Practical ACID example: Bank transfer
CREATE OR REPLACE PROCEDURE transfer_money(
    p_from_account VARCHAR2,
    p_to_account VARCHAR2,
    p_amount NUMBER
) IS
    v_from_balance NUMBER;
BEGIN
    -- Start transaction (implicit)
    
    -- Check sufficient funds
    SELECT balance INTO v_from_balance
    FROM accounts
    WHERE account_id = p_from_account
    FOR UPDATE; -- Lock the row
    
    IF v_from_balance < p_amount THEN
        RAISE_APPLICATION_ERROR(-20002, 'Insufficient funds');
    END IF;
    
    -- Perform transfer (atomicity)
    UPDATE accounts 
    SET balance = balance - p_amount 
    WHERE account_id = p_from_account;
    
    UPDATE accounts 
    SET balance = balance + p_amount 
    WHERE account_id = p_to_account;
    
    -- Log transaction
    INSERT INTO transaction_log (from_account, to_account, amount, trans_date)
    VALUES (p_from_account, p_to_account, p_amount, SYSDATE);
    
    COMMIT; -- Ensure durability
    
EXCEPTION
    WHEN OTHERS THEN
        ROLLBACK; -- Ensure atomicity
        RAISE;
END;
/

Before Runninginitial state

Before

accounts

account_id	owner	balance
1001	Alice	5,000
1002	Bob	3,200
1003	Carol	8,750

Data state before query executes.

Output / Afterresult

After

PROPERTY	GUARANTEED_BY	EXAMPLE
Atomicity	Transaction rollback	Transfer: both debit+credit or neither
Consistency	Constraints + triggers	Balance never goes negative
Isolation	Locking / MVCC	Session 2 sees no partial changes
Durability	Redo logs	Committed data survives power failure

ACID properties are the foundation of reliable relational database transactions

Key Takeaway

ACID stands for Atomicity, Consistency, Isolation, and Durability.

Q72. How do you handle transactions and locking in Oracle?

Answer

Oracle uses automatic transaction management with COMMIT/ROLLBACK and provides various locking mechanisms to ensure data consistency.

Sample Dataused in example

accounts

account_id	owner	balance
1001	Alice	5,000
1002	Bob	3,200
1003	Carol	8,750

CodeOracle SQL

SQL

-- Basic transaction control
BEGIN
   -- Start implicit transaction
   UPDATE employees SET salary = salary * 1.1 WHERE department_id = 10;
   INSERT INTO salary_audit (employee_id, old_salary, new_salary, change_date)
   SELECT employee_id, salary/1.1, salary, SYSDATE 
   FROM employees WHERE department_id = 10;
   
   -- Commit all changes
   COMMIT;
EXCEPTION
   WHEN OTHERS THEN
       -- Rollback on error
       ROLLBACK;
       RAISE;
END;
/

-- Savepoints for partial rollback
BEGIN
   UPDATE employees SET salary = salary + 1000 WHERE employee_id = 101;
   SAVEPOINT after_first_update;
   
   UPDATE employees SET salary = salary + 500 WHERE employee_id = 102;
   SAVEPOINT after_second_update;
   
   -- Some error condition
   IF SQL%ROWCOUNT = 0 THEN
       ROLLBACK TO after_first_update; -- Partial rollback
   END IF;
   
   COMMIT;
END;
/

-- Explicit locking with SELECT FOR UPDATE
-- Lock specific rows for update
SELECT employee_id, salary
FROM employees
WHERE department_id = 10
FOR UPDATE;

-- Update the locked rows
UPDATE employees 
SET salary = salary * 1.1 
WHERE department_id = 10;

COMMIT; -- Releases locks

-- Row-level locking with NOWAIT
BEGIN
   SELECT employee_id, salary
   FROM employees
   WHERE employee_id = 101
   FOR UPDATE NOWAIT; -- Don't wait if row is locked
   
   -- Perform updates
   UPDATE employees SET salary = 75000 WHERE employee_id = 101;
   COMMIT;
EXCEPTION
   WHEN RESOURCE_BUSY THEN
       DBMS_OUTPUT.PUT_LINE('Row is locked by another session');
END;
/

-- Lock specific columns
SELECT employee_id, salary, commission_pct
FROM employees
WHERE department_id = 20
FOR UPDATE OF salary; -- Only lock salary column

-- Check current locks
SELECT s.sid, s.serial#, s.username, s.status,
      l.type, l.mode_held, l.mode_requested,
      o.object_name
FROM v$session s
JOIN v$lock l ON s.sid = l.sid
JOIN dba_objects o ON l.id1 = o.object_id
WHERE s.username IS NOT NULL;

-- Deadlock handling
CREATE OR REPLACE PROCEDURE transfer_funds(
   p_from_account NUMBER,
   p_to_account NUMBER,
   p_amount NUMBER
) IS
   deadlock_detected EXCEPTION;
   PRAGMA EXCEPTION_INIT(deadlock_detected, -60);
BEGIN
   -- Lock accounts in consistent order to avoid deadlocks
   IF p_from_account < p_to_account THEN
       SELECT balance INTO l_from_balance
       FROM accounts WHERE account_id = p_from_account FOR UPDATE;
       
       SELECT balance INTO l_to_balance  
       FROM accounts WHERE account_id = p_to_account FOR UPDATE;
   ELSE
       SELECT balance INTO l_to_balance
       FROM accounts WHERE account_id = p_to_account FOR UPDATE;
       
       SELECT balance INTO l_from_balance
       FROM accounts WHERE account_id = p_from_account FOR UPDATE;
   END IF;
   
   -- Perform transfer
   UPDATE accounts SET balance = balance - p_amount WHERE account_id = p_from_account;
   UPDATE accounts SET balance = balance + p_amount WHERE account_id = p_to_account;
   
   COMMIT;
EXCEPTION
   WHEN deadlock_detected THEN
       ROLLBACK;
       DBMS_OUTPUT.PUT_LINE('Deadlock detected, transaction rolled back');
       RAISE;
END;
/

-- Set transaction isolation level
SET TRANSACTION ISOLATION LEVEL SERIALIZABLE;
-- or
SET TRANSACTION ISOLATION LEVEL READ COMMITTED;

-- Read-only transactions
SET TRANSACTION READ ONLY;
SELECT * FROM employees WHERE hire_date >= DATE '2023-01-01';
COMMIT;

-- Check transaction status
SELECT s.sid, s.serial#, s.username, 
      t.status, t.start_time, t.used_ublk
FROM v$session s
JOIN v$transaction t ON s.saddr = t.ses_addr
WHERE s.username = USER;

Before Runninginitial state

Before

accounts

account_id	owner	balance
1001	Alice	5,000
1002	Bob	3,200
1003	Carol	8,750

Data state before query executes.

Output / Afterresult

After

EXCEPTION	WHEN_RAISED	HANDLED_BY	OUTCOME
NO_DATA_FOUND	SELECT INTO = 0 rows	WHEN NO_DATA_FOUND	Continue after message
TOO_MANY_ROWS	SELECT INTO > 1 row	WHEN OTHERS	Rollback + log

Exceptions prevent unhandled errors from crashing the PL/SQL block

Key Takeaway

Oracle uses automatic transaction management with COMMIT/ROLLBACK and provides various locking mechanisms to ensure data consistency.

Q73. What do COMMIT, ROLLBACK, and SAVEPOINT do in a transaction?

Window / Analytic Functions

Q74. Explain window functions in SQL and provide examples.

Answer

Window functions in SQL perform calculations across a set of table rows related to the current row.

These functions allow users to carry out tasks like calculating running totals, ranking results, or computing averages within a specific window of data.

Window functions operate within a frame or window of data specified by the OVER clause.One common example is the ROW_NUMBER() function, which assigns a unique sequential integer to rows within a partition of a result set.

For exampleSELECT ROW_NUMBER() OVER (ORDER BY column_name) FROM table_name;assigns a unique number to each row ordered bycolumn_name.

Another example is the SUM() function used as a window function.

It calculates the cumulative sum of a column.SELECT SUM(column_name) OVER (ORDER BY column_name) FROM table_name;computes the running total ofcolumn_name.Window functions also include ranking functions like RANK() and DENSE_RANK().

These functions assign a rank to each row within a partition, with gaps in the rank values for ties in the case of RANK() and without gaps for DENSE_RANK().

For example,SELECT RANK() OVER (ORDER BY column_name) FROM table_name;ranks rows based oncolumn_name.These functions enhance the ability to perform complex calculations and data analysis directly within SQL queries, without requiring additional processing outside the database environment.

Their use is essential in scenarios where relational data needs advanced analytical processing like reporting, data analytics, or business intelligence applications.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

CodeOracle SQL

SQL

SELECT employee_name, salary,
  ROW_NUMBER()  OVER (PARTITION BY department_id ORDER BY salary DESC) AS row_num,
  RANK()        OVER (PARTITION BY department_id ORDER BY salary DESC) AS rnk,
  DENSE_RANK()  OVER (PARTITION BY department_id ORDER BY salary DESC) AS d_rnk
FROM employees;

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

Data state before query executes.

Output / Afterresult

After

EMP_NAME	SALARY	ROW_NUM	RANK	DENSE_RANK
Dave Brown	70,000	1	1	1
Bob Smith	62,000	2	2	2
Carol White	55,000	3	3	3
Alice Johnson	50,000	4	4	4

All 4 employees ranked by salary DESC — no ties so all methods agree

Key Takeaway

Window functions in SQL perform calculations across a set of table rows related to the current row.

Q75. Explain the use of rank, dense_rank, and row_number functions.

Answer

rank(), dense_rank(), and row_number() are SQL functions that enable you to assign ranks and numbers to rows within a result set, each with its unique behavior.

They are powerful tools for data analysis and reporting in SQL queries.Therank()function is used to assign a unique rank to each row within the result set based on the values in one or more columns.

It assigns the same rank to rows with equal values and leaves gaps in ranking for tied rows.

For example, rows with the same score will have the same rank, if you use rank() to rank a set of scores, and the next rank will be skipped.Thedense_rank()function is similar to rank() but does not leave gaps in ranking for tied rows.

It assigns the same rank to rows with equal values and continues with the next rank without gaps.

This is useful when you want a dense ranking without skipped ranks.Therow_number()function assigns a unique number to each row within the result set, without considering ties.

It simply numbers the rows sequentially, starting from 1.

This function is often used to generate a unique identifier for each row in a result set.These functions are valuable in various scenarios.

For example, in a competition where you want to rank participants based on their scores, use rank() to assign unique ranks while considering ties.row_number() is the choice, if you need a simple numbering of rows.

Dense_rank() is handy when you want dense ranking, such as in percentile calculations.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

CodeOracle SQL

SQL

SELECT employee_name, salary,
  ROW_NUMBER()  OVER (PARTITION BY department_id ORDER BY salary DESC) AS row_num,
  RANK()        OVER (PARTITION BY department_id ORDER BY salary DESC) AS rnk,
  DENSE_RANK()  OVER (PARTITION BY department_id ORDER BY salary DESC) AS d_rnk
FROM employees;

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

Data state before query executes.

Output / Afterresult

After

EMP_NAME	SALARY	ROW_NUM	RANK	DENSE_RANK
Dave Brown	70,000	1	1	1
Bob Smith	62,000	2	2	2
Carol White	55,000	3	3	3
Alice Johnson	50,000	4	4	4

All 4 employees ranked by salary DESC — no ties so all methods agree

Key Takeaway

rank(), dense_rank(), and row_number() are SQL functions that enable you to assign ranks and numbers to rows within a result set, each with its unique behavior.

Q76. Discuss the use of pivot and unpivot operations in SQL.

Answer

Pivot and unpivot operations in SQL are powerful tools for restructuring data within a database.

The pivot operation allows you to transform rows of data into columns, making it easier to analyze and present information.

The unpivot operation does the opposite, converting columns into rows.

These operations are particularly useful when dealing with complex datasets and when you need to transpose data to meet specific reporting or analysis requirements.Pivot operations are employed when you want to aggregate and summarize data by grouping it based on a particular column's values.

For example, A table is pivoted to display sales data by different product categories as separate columns, simplifying the process of comparing sales across categories.

This operation is especially valuable in scenarios where you need to generate cross-tabular reports or create visualizations that rely on data in a columnar format.Unpivot operations are beneficial when you have data stored in a wide format (columns) and need to convert it into a long format (rows).

This is commonly used in scenarios where data needs to be normalized for further analysis.

For example, unpivot it to transform it into a format where each row contains the month and corresponding sales value, if you have a table with columns representing different months of the year and sales data.

This makes it easier to perform time-series analysis and other operations.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

CodeOracle SQL

SQL

SELECT * FROM (
  SELECT department_id, EXTRACT(YEAR FROM hire_date) yr, salary
  FROM   employees
)
PIVOT (COUNT(*) FOR yr IN (2020, 2021, 2022, 2023));

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

Data state before query executes.

Output / Afterresult

After

DEPT_ID	2021_HIRES	2022_HIRES	2023_HIRES
10	1	1	0
20	1	0	1

PIVOT rotates row values into column headers — one column per year

Key Takeaway

Pivot and unpivot operations in SQL are powerful tools for restructuring data within a database.

Q77. How do you use window functions for running totals and moving averages in SQL?

Answer

Window functions are utilized for calculating running totals and moving averages.

TheSUM()function combined with theOVER()clause effectively computes running totals.

This approach involves specifying a window frame using theROWS BETWEENstatement, which defines the range of rows used in each calculation.

For example, to calculate a running total, useSUM(column_name) OVER (ORDER BY column_name).Similarly, for moving averages, theAVG()function is employed alongside theOVER()clause.

This function calculates the average over a specified range of rows, making it ideal for moving averages.

The syntaxAVG(column_name) OVER (ORDER BY column_name ROWS BETWEEN x PRECEDING AND y FOLLOWING)provides the average of values in a window frame defined by x preceding and y following rows.

This method is crucial in time series analysis and trend detection in datasets.

Employ these functions to accurately perform complex calculations in SQL, ensuring efficient and precise data analysis.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

CodeOracle SQL

SQL

SELECT employee_name, salary,
  ROW_NUMBER()  OVER (PARTITION BY department_id ORDER BY salary DESC) AS row_num,
  RANK()        OVER (PARTITION BY department_id ORDER BY salary DESC) AS rnk,
  DENSE_RANK()  OVER (PARTITION BY department_id ORDER BY salary DESC) AS d_rnk
FROM employees;

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

Data state before query executes.

Output / Afterresult

After

EMP_NAME	SALARY	ROW_NUM	RANK	DENSE_RANK
Dave Brown	70,000	1	1	1
Bob Smith	62,000	2	2	2
Carol White	55,000	3	3	3
Alice Johnson	50,000	4	4	4

All 4 employees ranked by salary DESC — no ties so all methods agree

Key Takeaway

Window functions are utilized for calculating running totals and moving averages.

Q78. Explain the difference between RANK(), DENSE_RANK(), and ROW_NUMBER() with practical examples.

Answer

These ranking functions handle ties differently: ROW_NUMBER() assigns unique sequential numbers, RANK() skips ranks after ties, DENSE_RANK() assigns consecutive ranks without gaps.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

CodeOracle SQL

SQL

-- Sample data with ties
WITH sample_scores AS (
   SELECT 'Alice' as name, 95 as score FROM dual UNION ALL
   SELECT 'Bob', 87 FROM dual UNION ALL
   SELECT 'Carol', 95 FROM dual UNION ALL
   SELECT 'David', 82 FROM dual UNION ALL
   SELECT 'Eve', 87 FROM dual
)
SELECT name, score,
      ROW_NUMBER() OVER (ORDER BY score DESC) as row_num,
      RANK() OVER (ORDER BY score DESC) as rank_val,
      DENSE_RANK() OVER (ORDER BY score DESC) as dense_rank_val
FROM sample_scores;

-- Results:
-- Alice    95    1    1    1
-- Carol    95    2    1    1  (tie handling difference)
-- Bob      87    3    3    2  (RANK skips 2, DENSE_RANK doesn't)
-- Eve      87    4    3    2
-- David    82    5    5    3

-- Practical use case: Top N per group
SELECT department_id, employee_id, salary, salary_rank
FROM (
   SELECT department_id, employee_id, salary,
          DENSE_RANK() OVER (PARTITION BY department_id ORDER BY salary DESC) as salary_rank
   FROM employees
)
WHERE salary_rank <= 3;

-- When to use which:
-- ROW_NUMBER(): When you need unique ranking (pagination, unique identifiers)
-- RANK(): When gaps in ranking are acceptable (sports rankings)
-- DENSE_RANK(): When you need continuous ranking (grade classifications)

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

Data state before query executes.

Output / Afterresult

After

EMP_NAME	SALARY	ROW_NUM	RANK	DENSE_RANK
Dave Brown	70,000	1	1	1
Bob Smith	62,000	2	2	2
Carol White	55,000	3	3	3
Alice Johnson	50,000	4	4	4

All 4 employees ranked by salary DESC — no ties so all methods agree

Key Takeaway

These ranking functions handle ties differently: ROW_NUMBER() assigns unique sequential numbers, RANK() skips ranks after ties, DENSE_RANK() assigns consecutive ranks without gaps.

Q79. How do you calculate running totals and moving averages using window functions?

Answer

Use window functions with frame specifications (ROWS/RANGE BETWEEN) to calculate cumulative values and moving averages over ordered datasets.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

CodeOracle SQL

SQL

-- Running totals and moving averages
SELECT month_date, sales_amount,
      -- Running total (cumulative sum)
      SUM(sales_amount) OVER (ORDER BY month_date 
                             ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW) as running_total,
      
      -- 3-month moving average
      AVG(sales_amount) OVER (ORDER BY month_date 
                             ROWS BETWEEN 2 PRECEDING AND CURRENT ROW) as moving_avg_3m,
      
      -- 6-month moving total
      SUM(sales_amount) OVER (ORDER BY month_date 
                             ROWS BETWEEN 5 PRECEDING AND CURRENT ROW) as moving_total_6m,
      
      -- Year-to-date calculation
      SUM(sales_amount) OVER (PARTITION BY EXTRACT(YEAR FROM month_date) 
                             ORDER BY month_date) as ytd_sales,
      
      -- Percentage of running total
      ROUND(sales_amount / SUM(sales_amount) OVER (ORDER BY month_date 
                                                  ROWS UNBOUNDED PRECEDING) * 100, 2) as pct_running_total
FROM monthly_sales
ORDER BY month_date;

-- Advanced frame specifications
SELECT employee_id, salary, hire_date,
      -- Centered moving average (previous, current, next)
      AVG(salary) OVER (ORDER BY hire_date 
                       ROWS BETWEEN 1 PRECEDING AND 1 FOLLOWING) as centered_avg,
      
      -- Range-based frame (salary within 5000 range)
      AVG(salary) OVER (ORDER BY salary 
                       RANGE BETWEEN 2500 PRECEDING AND 2500 FOLLOWING) as range_avg
FROM employees;

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

Data state before query executes.

Output / Afterresult

After

EMP_NAME	SALARY	RUNNING_TOTAL
Alice Johnson	50,000	50,000
Carol White	55,000	105,000
Bob Smith	62,000	167,000
Dave Brown	70,000	237,000

Running total — cumulative SUM() OVER (ORDER BY salary ASC)

Key Takeaway

Use window functions with frame specifications (ROWS/RANGE BETWEEN) to calculate cumulative values and moving averages over ordered datasets.

Q80. How do you pivot and unpivot data in Oracle SQL?

Q81. What do the LAG() and LEAD() analytic functions do?

Query Performance & Optimization

Q82. How do you optimize an SQL query?

Q83. How would you approach performance tuning in a SQL database?

Q84. How does a database optimizer work?

Q85. Describe the process and considerations for SQL query optimization in large datasets.

Answer

Understanding the data structure and indexing is crucial, in SQL query optimization for large datasets.

Efficient indexing enhances query performance by reducing the search space.

Analyze query execution plans to identify bottlenecks, as they reveal the steps the database takes to execute a query.

Use joins appropriately; overusing them can lead to slow performance.

Optimize joins by ensuring that the join conditions are on indexed columns.

Aggregate functions should be used judiciously.

They can slow down queries if applied to large data sets without proper filtering.

Ensure that WHERE clauses are selective, as this filters out unnecessary data early in the query execution process.Proper use of subqueries and temporary tables can significantly improve performance.

Subqueries should be used for complex filtering and aggregation, but avoid them in cases where joins can be more efficient.

Temporary tables are useful for breaking down complex queries into simpler steps.

This approach can improve readability and performance.

Regularly update statistics on the database to help the query optimizer make better decisions.

This is especially important in dynamic environments where data changes frequently.

Lastly, avoid using functions on indexed columns in the WHERE clause.

This practice can prevent the database from using the index, leading to full table scans and slower performance.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

CodeOracle SQL

SQL

SELECT * FROM employees WHERE ROWNUM <= 5;

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

Data state before query executes.

Output / Afterresult

After

EMP_ID	EMP_NAME	SALARY
101	Alice Johnson	50,000
102	Bob Smith	62,000
103	Carol White	55,000

Top 3 rows — ROWNUM/FETCH FIRST limits result set

Key Takeaway

Understanding the data structure and indexing is crucial, in SQL query optimization for large datasets.

Q86. How do you handle large-scale data deletion without impacting database performance?

Q87. How do you approach capacity planning for a SQL database?

Q88. What are the advanced techniques for SQL query performance tuning?

Q89. How can you optimize an Oracle SQL query for better performance?

Answer

Query optimization in Oracle involves proper indexing, execution plan analysis, hint usage, and understanding the cost-based optimizer.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

CodeOracle SQL

SQL

-- 1. Use proper indexes
CREATE INDEX idx_emp_dept_salary ON employees(department_id, salary);

-- 2. Check execution plan
EXPLAIN PLAN FOR
SELECT * FROM employees WHERE department_id = 10 AND salary > 50000;

SELECT * FROM TABLE(DBMS_XPLAN.DISPLAY);

-- 3. Use bind variables instead of literals
SELECT * FROM employees WHERE employee_id = :emp_id;

-- 4. Avoid SELECT *
SELECT employee_id, first_name, salary 
FROM employees 
WHERE department_id = 10;

-- 5. Use EXISTS instead of IN for subqueries
SELECT * FROM departments d
WHERE EXISTS (SELECT 1 FROM employees e WHERE e.department_id = d.department_id);

-- 6. Use analytical functions instead of self-joins
SELECT employee_id, salary,
       RANK() OVER (ORDER BY salary DESC) as salary_rank
FROM employees;

-- 7. Use ROWNUM for limiting results (Oracle-specific)
SELECT * FROM (
    SELECT * FROM employees ORDER BY salary DESC
) WHERE ROWNUM <= 10;

-- 8. Oracle 12c+ OFFSET FETCH syntax
SELECT * FROM employees
ORDER BY salary DESC
OFFSET 0 ROWS FETCH NEXT 10 ROWS ONLY;

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

Data state before query executes.

Output / Afterresult

After

EMP_NAME	SALARY	ROW_NUM	RANK	DENSE_RANK
Dave Brown	70,000	1	1	1
Bob Smith	62,000	2	2	2
Carol White	55,000	3	3	3
Alice Johnson	50,000	4	4	4

All 4 employees ranked by salary DESC — no ties so all methods agree

Key Takeaway

Query optimization in Oracle involves proper indexing, execution plan analysis, hint usage, and understanding the cost-based optimizer.

Q90. How can you improve the performance of an Oracle database?

Answer

Oracle database performance can be improved through proper indexing, query optimization, database tuning, memory management, and hardware optimization.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

CodeOracle SQL

SQL

-- 1. INDEXING STRATEGIES
-- Create appropriate indexes
CREATE INDEX idx_emp_dept_join ON employees(department_id);
CREATE INDEX idx_emp_salary_range ON employees(salary);

-- Composite indexes for multi-column queries
CREATE INDEX idx_emp_dept_salary_name ON employees(department_id, salary, last_name);

-- Function-based indexes for case-insensitive searches
CREATE INDEX idx_emp_upper_name ON employees(UPPER(last_name));

-- 2. QUERY OPTIMIZATION
-- Use bind variables
SELECT * FROM employees WHERE employee_id = :emp_id;

-- Avoid SELECT *
SELECT employee_id, first_name, salary FROM employees;

-- Use EXISTS instead of IN
SELECT * FROM departments d
WHERE EXISTS (SELECT 1 FROM employees e WHERE e.department_id = d.department_id);

-- 3. STATISTICS COLLECTION
-- Gather table statistics
BEGIN
    DBMS_STATS.GATHER_TABLE_STATS('HR', 'EMPLOYEES');
END;
/

-- Gather schema statistics
BEGIN
    DBMS_STATS.GATHER_SCHEMA_STATS('HR');
END;
/

-- 4. MEMORY TUNING
-- Check SGA components
SELECT component, current_size/1024/1024 as size_mb
FROM v$sga_dynamic_components;

-- Tune buffer cache hit ratio
SELECT name, value
FROM v$sysstat
WHERE name IN ('db block gets', 'consistent gets', 'physical reads');

-- 5. SQL TUNING ADVISOR
-- Create tuning task
DECLARE
    task_name VARCHAR2(30);
BEGIN
    task_name := DBMS_SQLTUNE.CREATE_TUNING_TASK(
        sql_text => 'SELECT * FROM employees WHERE department_id = 10'
    );
    DBMS_SQLTUNE.EXECUTE_TUNING_TASK(task_name);
END;
/

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

Data state before query executes.

Output / Afterresult

After

EMP_ID	EMP_NAME	DEPT_ID	SALARY
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	20	70,000

All 4 rows returned

Key Takeaway

Oracle database performance can be improved through proper indexing, query optimization, database tuning, memory management, and hardware optimization.

Q91. What is EXPLAIN PLAN and how is it used to analyze a query?

Database Design & Normalization

Q92. What is normalization?

Q93. What are the different normal forms in SQL?

Answer

Different normal forms in SQL are levels of organizing and structuring relational databases to minimize data redundancy and improve data integrity.

These normal forms, denoted as 1NF, 2NF, 3NF, BCNF, and 4NF, each have specific rules and conditions that a database table must satisfy.1NF, or First Normal Form, ensures that each column in a table contains atomic (indivisible) values, and there are no repeating groups or arrays.2NF, or Second Normal Form, builds upon 1NF and requires that non-key attributes are functionally dependent on the entire primary key.3NF, or Third Normal Form, extends the normalization process by eliminating transitive dependencies between non-key attributes.BCNF, or Boyce-Codd Normal Form, deals with situations where a table has multiple candidate keys and ensures that non-prime attributes are functionally dependent only on the candidate keys.4NF, or Fourth Normal Form, addresses multivalued dependencies in a table, ensuring that they are properly handled.These normal forms help in designing efficient and well-structured databases, minimizing data anomalies, and ensuring data integrity, which are essential principles in SQL database design.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

CodeOracle SQL

SQL

SELECT * FROM employees WHERE ROWNUM <= 5;

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

Data state before query executes.

Output / Afterresult

After

NORMAL_FORM	RULE_VIOLATED	FIX
0NF → 1NF	Repeating groups: phone='123,456'	Separate into individual rows
1NF → 2NF	Partial dependency on composite PK	Move to separate table
2NF → 3NF	Transitive: emp→dept_id→dept_name	Move dept_name to departments table
3NF → BCNF	Determinant not a candidate key	Decompose further

Normalize step by step — each form removes a specific type of redundancy

Key Takeaway

Different normal forms in SQL are levels of organizing and structuring relational databases to minimize data redundancy and improve data integrity.

Q94. What is denormalization and when would you use it?

Q95. Explain the concept of database normalization beyond the third normal form.

Q96. Describe the process of data modeling for a complex business process.

Answer

The process of data modeling for a complex business process involves several key steps.

It requires understanding and defining the business requirements, which involves identifying the data entities and their relationships.

This step often includes extensive discussions with stakeholders to ensure all business needs are captured.

The next phase is the creation of an Entity-Relationship (ER) diagram.

This diagram visually represents the data entities, attributes, and the relationships between them, providing a clear structure for the database.Once the ER diagram is complete, the next step is to translate it into a database schema using SQL.

This involves defining tables, primary and secondary keys, and establishing relationships through foreign keys.

The schema must be normalized to eliminate data redundancy and improve data integrity.

Normalization typically involves organizing the data into tables and establishing relationships between these tables based on rules designed to protect the data and make the database more efficient.

The database is ready after normalization ,for implementation and further optimization, if necessary, to support complex business queries and operations.

This final step ensures the database is not only structured according to business requirements but also performs efficiently under various data operations.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

CodeOracle SQL

SQL

SELECT * FROM employees WHERE ROWNUM <= 5;

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

Data state before query executes.

Output / Afterresult

After

EMP_ID	EMP_NAME	SALARY
101	Alice Johnson	50,000
102	Bob Smith	62,000
103	Carol White	55,000

Top 3 rows — ROWNUM/FETCH FIRST limits result set

Key Takeaway

The process of data modeling for a complex business process involves several key steps.

Q97. Describe the process of data denormalization and its impact on performance and scalability.

Q98. Explain the difference between Oracle storage engines and database architecture.

Answer

Unlike MySQL with MyISAM/InnoDB, Oracle uses a unified architecture with tablespaces, data files, and advanced features like Real Application Clusters (RAC).

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

CodeOracle SQL

SQL

-- Oracle Database Architecture Components

-- 1. TABLESPACES - Logical storage units
CREATE TABLESPACE hr_data
DATAFILE '/u01/app/oracle/oradata/hr_data01.dbf' SIZE 100M
AUTOEXTEND ON NEXT 10M MAXSIZE 1G;

-- Create table in specific tablespace
CREATE TABLE employees (
    employee_id NUMBER PRIMARY KEY,
    first_name VARCHAR2(50),
    last_name VARCHAR2(50),
    salary NUMBER
) TABLESPACE hr_data;

-- 2. SCHEMAS - Logical container for database objects
CREATE USER hr_user IDENTIFIED BY password
DEFAULT TABLESPACE hr_data
TEMPORARY TABLESPACE temp;

GRANT CREATE SESSION, CREATE TABLE TO hr_user;

-- 3. SEGMENTS, EXTENTS, AND BLOCKS
-- Check segment information
SELECT segment_name, segment_type, tablespace_name, bytes/1024/1024 as size_mb
FROM user_segments
WHERE segment_name = 'EMPLOYEES';

-- Check extent information
SELECT extent_id, bytes/1024 as size_kb, blocks
FROM user_extents
WHERE segment_name = 'EMPLOYEES'
ORDER BY extent_id;

-- 4. ORACLE MEMORY STRUCTURES
-- System Global Area (SGA)
SELECT component, current_size/1024/1024 as size_mb
FROM v$sga_dynamic_components;

-- Buffer Cache
SELECT name, value
FROM v$sysstat
WHERE name IN ('db block gets', 'consistent gets', 'physical reads');

-- 5. ORACLE PROCESSES
-- Check background processes
SELECT pname, description
FROM v$bgprocess
WHERE pname IS NOT NULL;

-- 6. REDO LOGS AND ARCHIVE LOGS
SELECT group#, status, archived, bytes/1024/1024 as size_mb
FROM v$log;

-- Archive log mode check
SELECT log_mode FROM v$database;

-- 7. DATA DICTIONARY VIEWS
-- Table information
SELECT table_name, tablespace_name, num_rows, blocks
FROM user_tables
WHERE table_name = 'EMPLOYEES';

-- Index information
SELECT index_name, table_name, uniqueness, status
FROM user_indexes
WHERE table_name = 'EMPLOYEES';

-- 8. ORACLE ADVANCED FEATURES
-- Partitioning
CREATE TABLE sales_partitioned (
    sale_id NUMBER,
    sale_date DATE,
    amount NUMBER
)
PARTITION BY RANGE (sale_date) (
    PARTITION p_2023 VALUES LESS THAN (DATE '2024-01-01'),
    PARTITION p_2024 VALUES LESS THAN (DATE '2025-01-01')
);

-- Materialized Views
CREATE MATERIALIZED VIEW mv_sales_summary
BUILD IMMEDIATE
REFRESH FAST ON COMMIT
AS
SELECT TRUNC(sale_date, 'MM') as sale_month,
       SUM(amount) as total_sales
FROM sales
GROUP BY TRUNC(sale_date, 'MM');

-- 9. ORACLE RAC (Real Application Clusters)
-- Check RAC configuration
SELECT instance_name, host_name, status
FROM gv$instance;

-- 10. AUTOMATIC STORAGE MANAGEMENT (ASM)
-- Check ASM disk groups
SELECT name, state, type, total_mb, free_mb
FROM v$asm_diskgroup;

-- Oracle vs Other Databases:
-- - Unified architecture (no separate storage engines)
-- - Advanced clustering with RAC
-- - Automatic Storage Management (ASM)
-- - Advanced partitioning options
-- - Flashback technology
-- - Advanced security features

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

Data state before query executes.

Output / Afterresult

After

DEPT_ID	EMP_COUNT	AVG_SALARY	MAX_SALARY
10	2	52,500	55,000
20	2	66,000	70,000

2 groups — Alice+Carol in dept 10, Bob+Dave in dept 20

Key Takeaway

Unlike MySQL with MyISAM/InnoDB, Oracle uses a unified architecture with tablespaces, data files, and advanced features like Real Application Clusters (RAC).

Data Warehousing, ETL & BI

Q99. What is a data warehouse?

Q100. What is data mining?

Q101. Describe the difference between OLAP and OLTP databases.

Answer

OLAP databases often use a star schema or snowflake schema for organizing data, which aids in complex querying and analysis.

OLTP databases typically utilize a normalized schema to optimize for transactional speed and data integrity.

While OLAP databases facilitate strategic business decisions through data analysis, OLTP systems focus on the efficient processing of day-to-day transactions.The difference between OLAP and OLTP databases lies in their primary functions and design structures.

OLAP (Online Analytical Processing) databases are optimized for complex query processing and data analysis.

They support decision-making processes and data discovery in business environments.

OLAP systems work efficiently with large data sets, enabling multidimensional queries, often in a data warehouse setting.

These databases are designed for read-intensive operations and facilitate data aggregation, trend analysis, and complex calculations.OLTP (Online Transaction Processing) databases are tailored for managing transaction-oriented applications.

OLTP is adept at handling a large number of short, atomic transactions such as insert, update, and delete operations.

OLTP systems are characterized by their ability to maintain data integrity in multi-access environments.

They are optimized for speed and efficiency in transaction processing, ensuring rapid response times and high throughput.

OLTP databases are commonly used in retail, banking, and other industries where transactional data is constantly processed.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

CodeOracle SQL

SQL

SELECT * FROM employees WHERE ROWNUM <= 5;

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

Data state before query executes.

Output / Afterresult

After

EMP_ID	EMP_NAME	SALARY
101	Alice Johnson	50,000
102	Bob Smith	62,000
103	Carol White	55,000

Top 3 rows — ROWNUM/FETCH FIRST limits result set

Key Takeaway

OLAP databases often use a star schema or snowflake schema for organizing data, which aids in complex querying and analysis.

Q102. Describe the process of data warehousing.

Q103. What is the role of ETL (Extract, Transform, Load) in data processing?

Answer

ETL (Extract, Transform, Load) plays a crucial role in data processing within the realm of SQL and databases.

ETL serves as the backbone for moving and manipulating data to ensure it is ready for analysis and reporting."Extract," data is gathered from various sources, such as databases, spreadsheets, or external systems, in the first phase.

SQL queries are commonly used to extract relevant data subsets.

This step ensures that the required data is accessible for further processing.SQL is employed to clean, reshape, and enrich the extracted data, in the "Transform" phase.

This involves tasks like data cleansing, validation, and aggregation.

SQL's powerful querying capabilities enable data engineers to perform complex transformations to meet specific business requirements.The transformed data is loaded into a target database or data warehouse, in the "Load" phase.

SQL's data loading capabilities facilitate this process, ensuring that the data is stored efficiently and is ready for analysis through SQL queries.The benefits of ETL in data processing are manifold.

It enables organizations to maintain clean and structured data, ensuring data accuracy.

ETL also allows for the integration of data from disparate sources into a unified format, enabling comprehensive analysis.

SQL's role in this process is pivotal, as it provides a robust language for data manipulation and transformation, making ETL processes efficient and effective.

Overall, ETL, driven by SQL, streamlines data processing, making it a fundamental component of modern data-driven organizations.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

CodeOracle SQL

SQL

SELECT * FROM employees WHERE ROWNUM <= 5;

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

Data state before query executes.

Output / Afterresult

After

EMP_ID	EMP_NAME	SALARY
101	Alice Johnson	50,000
102	Bob Smith	62,000
103	Carol White	55,000

Top 3 rows — ROWNUM/FETCH FIRST limits result set

Key Takeaway

ETL (Extract, Transform, Load) plays a crucial role in data processing within the realm of SQL and databases.

Q104. Discuss the use of SQL in business intelligence and data visualization.

Security & SQL Injection

Q105. Explain what SQL Injection is and how to prevent it.

Q106. What is SQL injection and how do you prevent it in stored procedures?

Q107. What are the best practices for securing a SQL database?

Q108. What are the best practices for database encryption and securing sensitive data in SQL?

Q109. Explain the role of SQL in data governance and compliance.

Q110. How can you prevent SQL injection in Oracle applications?

Answer

Use bind variables (parameterized queries), validate input data, avoid dynamic SQL when possible, and use proper escaping techniques.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

CodeOracle SQL

SQL

-- 1. Use Bind Variables (Parameterized Queries)
-- BAD - Vulnerable to SQL injection
CREATE OR REPLACE PROCEDURE bad_login(p_username VARCHAR2, p_password VARCHAR2) IS
    v_sql VARCHAR2(1000);
    v_count NUMBER;
BEGIN
    v_sql := 'SELECT COUNT(*) FROM users WHERE username = ''' || 
             p_username || ''' AND password = ''' || p_password || '''';
    EXECUTE IMMEDIATE v_sql INTO v_count;
END;

-- GOOD - Using bind variables
CREATE OR REPLACE PROCEDURE safe_login(p_username VARCHAR2, p_password VARCHAR2) IS
    v_count NUMBER;
BEGIN
    SELECT COUNT(*) INTO v_count
    FROM users 
    WHERE username = p_username 
    AND password = p_password;
END;

-- 2. For dynamic SQL, use bind variables
CREATE OR REPLACE PROCEDURE safe_dynamic_query(
    p_table_name VARCHAR2,
    p_employee_id NUMBER
) IS
    v_sql VARCHAR2(1000);
    v_count NUMBER;
BEGIN
    -- Validate table name against whitelist
    IF p_table_name NOT IN ('employees', 'departments', 'jobs') THEN
        RAISE_APPLICATION_ERROR(-20001, 'Invalid table name');
    END IF;
    
    v_sql := 'SELECT COUNT(*) FROM ' || p_table_name || 
             ' WHERE employee_id = :1';
    EXECUTE IMMEDIATE v_sql INTO v_count USING p_employee_id;
END;

-- 3. Input validation function
CREATE OR REPLACE FUNCTION validate_input(p_input VARCHAR2) RETURN VARCHAR2 IS
BEGIN
    -- Remove dangerous characters
    RETURN REGEXP_REPLACE(p_input, '[''";\\-]', '');
END;

-- 4. Use DBMS_ASSERT for validation
CREATE OR REPLACE PROCEDURE validate_object_name(p_table_name VARCHAR2) IS
    v_valid_name VARCHAR2(128);
BEGIN
    -- This will raise an exception if not a valid SQL name
    v_valid_name := DBMS_ASSERT.SQL_OBJECT_NAME(p_table_name);
    
    -- Additional whitelist check
    IF v_valid_name NOT IN ('EMPLOYEES', 'DEPARTMENTS') THEN
        RAISE_APPLICATION_ERROR(-20002, 'Table not allowed');
    END IF;
END;

-- 5. Safe cursor with bind variables
CREATE OR REPLACE PROCEDURE search_employees(
    p_search_term VARCHAR2,
    p_cursor OUT SYS_REFCURSOR
) IS
BEGIN
    OPEN p_cursor FOR
        SELECT employee_id, first_name, last_name
        FROM employees
        WHERE UPPER(first_name) LIKE UPPER('%' || p_search_term || '%')
           OR UPPER(last_name) LIKE UPPER('%' || p_search_term || '%');
END;

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

Data state before query executes.

Output / Afterresult

After

APPROACH	INPUT	SAFE	RESULT
String concat	' OR '1'='1	❌ No	Returns ALL 4 rows — security breach
Bind variable :p1	' OR '1'='1	✅ Yes	0 rows — treated as literal data

Bind variables prevent SQL injection — the #1 Oracle security best practice

Key Takeaway

Use bind variables (parameterized queries), validate input data, avoid dynamic SQL when possible, and use proper escaping techniques.

Q111. Write a query to implement row-level security (RLS) in Oracle.

Answer

Use Oracle Virtual Private Database (VPD) policies to implement row-level security that automatically filters data based on user context.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

CodeOracle SQL

SQL

-- Create policy function for row-level security
CREATE OR REPLACE FUNCTION employee_security_policy(
   schema_var IN VARCHAR2,
   table_var IN VARCHAR2
) RETURN VARCHAR2 IS
   v_user VARCHAR2(100);
   v_predicate VARCHAR2(4000);
BEGIN
   v_user := SYS_CONTEXT('USERENV', 'SESSION_USER');
   
   -- Different access levels based on user
   CASE 
       WHEN v_user IN ('HR_ADMIN', 'SYSTEM') THEN
           -- Full access
           v_predicate := '1=1';
       WHEN v_user LIKE 'MGR_%' THEN
           -- Managers can see their department employees
           v_predicate := 'department_id IN (
               SELECT department_id FROM departments 
               WHERE manager_id = (
                   SELECT employee_id FROM employees 
                   WHERE user_id = ''' || v_user || '''
               )
           )';
       WHEN v_user LIKE 'EMP_%' THEN
           -- Employees can only see their own record
           v_predicate := 'employee_id = (
               SELECT employee_id FROM employees 
               WHERE user_id = ''' || v_user || '''
           )';
       ELSE
           -- Default: no access
           v_predicate := '1=0';
   END CASE;
   
   RETURN v_predicate;
END;
/

-- Apply the policy to employees table
BEGIN
   DBMS_RLS.ADD_POLICY(
       object_schema => 'HR',
       object_name => 'EMPLOYEES',
       policy_name => 'EMPLOYEE_ACCESS_POLICY',
       function_schema => 'HR',
       policy_function => 'EMPLOYEE_SECURITY_POLICY',
       statement_types => 'SELECT, INSERT, UPDATE, DELETE',
       update_check => TRUE
   );
END;
/

-- Context-based security policy
CREATE OR REPLACE FUNCTION dept_security_policy(
   schema_var IN VARCHAR2,
   table_var IN VARCHAR2
) RETURN VARCHAR2 IS
   v_dept_id NUMBER;
   v_role VARCHAR2(100);
BEGIN
   -- Get user's department and role from application context
   v_dept_id := SYS_CONTEXT('USER_CONTEXT', 'DEPARTMENT_ID');
   v_role := SYS_CONTEXT('USER_CONTEXT', 'USER_ROLE');
   
   IF v_role = 'ADMIN' THEN
       RETURN '1=1'; -- Full access
   ELSIF v_role = 'MANAGER' THEN
       RETURN 'department_id = ' || v_dept_id;
   ELSIF v_role = 'EMPLOYEE' THEN
       RETURN 'employee_id = ' || SYS_CONTEXT('USER_CONTEXT', 'EMPLOYEE_ID');
   ELSE
       RETURN '1=0'; -- No access
   END IF;
END;
/

-- Create application context
CREATE OR REPLACE CONTEXT user_context USING security_pkg;

-- Package to set context
CREATE OR REPLACE PACKAGE security_pkg IS
   PROCEDURE set_user_context(
       p_user_id VARCHAR2,
       p_department_id NUMBER,
       p_employee_id NUMBER,
       p_role VARCHAR2
   );
END;
/

CREATE OR REPLACE PACKAGE BODY security_pkg IS
   PROCEDURE set_user_context(
       p_user_id VARCHAR2,
       p_department_id NUMBER,
       p_employee_id NUMBER,
       p_role VARCHAR2
   ) IS
   BEGIN
       DBMS_SESSION.SET_CONTEXT('USER_CONTEXT', 'USER_ID', p_user_id);
       DBMS_SESSION.SET_CONTEXT('USER_CONTEXT', 'DEPARTMENT_ID', p_department_id);
       DBMS_SESSION.SET_CONTEXT('USER_CONTEXT', 'EMPLOYEE_ID', p_employee_id);
       DBMS_SESSION.SET_CONTEXT('USER_CONTEXT', 'USER_ROLE', p_role);
   END;
END;
/

-- Column-level security policy
CREATE OR REPLACE FUNCTION salary_security_policy(
   schema_var IN VARCHAR2,
   table_var IN VARCHAR2
) RETURN VARCHAR2 IS
   v_role VARCHAR2(100);
BEGIN
   v_role := SYS_CONTEXT('USER_CONTEXT', 'USER_ROLE');
   
   -- Only managers and HR can see salaries
   IF v_role IN ('MANAGER', 'HR_ADMIN') THEN
       RETURN '1=1';
   ELSE
       RETURN '1=0';
   END IF;
END;
/

-- Apply column-level policy
BEGIN
   DBMS_RLS.ADD_POLICY(
       object_schema => 'HR',
       object_name => 'EMPLOYEES',
       policy_name => 'SALARY_COLUMN_POLICY',
       function_schema => 'HR',
       policy_function => 'SALARY_SECURITY_POLICY',
       statement_types => 'SELECT',
       sec_relevant_cols => 'SALARY,COMMISSION_PCT',
       sec_relevant_cols_opt => DBMS_RLS.ALL_ROWS
   );
END;
/

-- Dynamic policy based on time and IP
CREATE OR REPLACE FUNCTION time_ip_security_policy(
   schema_var IN VARCHAR2,
   table_var IN VARCHAR2
) RETURN VARCHAR2 IS
   v_current_hour NUMBER;
   v_ip_address VARCHAR2(100);
   v_allowed_networks SYS.ODCIVARCHAR2LIST;
BEGIN
   v_current_hour := EXTRACT(HOUR FROM SYSTIMESTAMP);
   v_ip_address := SYS_CONTEXT('USERENV', 'IP_ADDRESS');
   v_allowed_networks := SYS.ODCIVARCHAR2LIST('192.168.1.%', '10.0.0.%');
   
   -- Business hours check (9 AM to 6 PM)
   IF v_current_hour < 9 OR v_current_hour > 18 THEN
       RETURN '1=0'; -- No access outside business hours
   END IF;
   
   -- IP address check
   FOR i IN 1..v_allowed_networks.COUNT LOOP
       IF v_ip_address LIKE v_allowed_networks(i) THEN
           RETURN '1=1'; -- Allow access from trusted networks
       END IF;
   END LOOP;
   
   RETURN '1=0'; -- Default deny
END;
/

-- View current policies
SELECT object_owner, object_name, policy_name, function, 
      enable_flag, sel, ins, upd, del
FROM dba_policies
WHERE object_name = 'EMPLOYEES';

-- Test the security policy
-- Set context for a manager user
BEGIN
   security_pkg.set_user_context('MGR_001', 10, 100, 'MANAGER');
END;
/

-- This query will only return employees from department 10
SELECT employee_id, first_name, last_name, department_id
FROM employees;

-- Drop policy if needed
BEGIN
   DBMS_RLS.DROP_POLICY(
       object_schema => 'HR',
       object_name => 'EMPLOYEES',
       policy_name => 'EMPLOYEE_ACCESS_POLICY'
   );
END;
/

-- Audit policy usage
SELECT policy_name, sql_text, timestamp, username
FROM dba_fga_audit_trail
WHERE object_name = 'EMPLOYEES'
ORDER BY timestamp DESC;

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

Data state before query executes.

Output / Afterresult

After

EMPLOYEE	MANAGER
Bob Smith	Alice Johnson
Carol White	Alice Johnson
Dave Brown	Bob Smith

3 rows — self join pairs each employee with their manager's name

Key Takeaway

Use Oracle Virtual Private Database (VPD) policies to implement row-level security that automatically filters data based on user context.

Q112. What do GRANT and REVOKE do in Oracle's Data Control Language?

Distributed Systems, Replication & Sharding

Q113. What is database sharding and when is it useful?

Q114. Explain the concept of database partitioning and its types.

Q115. Describe the use and benefits of database replication.

Answer

Database replication in SQL ensures data availability, load balancing, disaster recovery, and scalability, making it a valuable strategy for maintaining robust and high-performing database systems.

Database replication in the context of SQL serves as a vital mechanism for duplicating and maintaining database content across multiple servers.

This process ensures data consistency, availability, and fault tolerance, benefiting SQL systems in several ways.Replication enhances data availability by creating redundant copies of the database on separate servers.

This redundancy enables continued access to data even in the event of server failures or maintenance, minimizing downtime.

It contributes to load balancing by distributing read queries among replicated servers, reducing the workload on the primary server.

This optimizes query performance, resulting in faster response times for users.Database replication supports disaster recovery.

In case of data corruption or loss, the replicated copies are used to restore the database to a previous state, safeguarding critical information.

Replication aids in scaling SQL systems horizontally.

Additional servers are added to the replication setup as data grows, accommodating increased data demands without significantly affecting performance.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

CodeOracle SQL

SQL

SELECT * FROM employees WHERE ROWNUM <= 5;

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

Data state before query executes.

Output / Afterresult

After

EMP_ID	EMP_NAME	SALARY
101	Alice Johnson	50,000
102	Bob Smith	62,000
103	Carol White	55,000

Top 3 rows — ROWNUM/FETCH FIRST limits result set

Key Takeaway

Database replication in SQL ensures data availability, load balancing, disaster recovery, and scalability, making it a valuable strategy for maintaining robust and high-performing database systems.

Q116. Discuss the techniques for ensuring data consistency across distributed databases in SQL.

Answer

Ensuring data consistency across distributed databases in SQL involves key techniques are distributed transactions, replication, conflict resolution and data versioning.

Use of distributed transactions guarantees that operations across multiple databases either all succeed or all fail, maintaining data integrity.

This is achieved through two-phase commit protocols, where the first phase prepares all databases for a transaction and the second phase either commits or rolls back the transaction based on a unanimous agreement among the databases.Another important technique is the implementation of replication, where changes made in one database are automatically mirrored in others.

This ensures that all distributed databases hold consistent and up-to-date data.

Conflict resolution strategies are crucial when simultaneous updates occur in different databases; these strategies determine which version of the data is considered authoritative.

Replication can be synchronous, where transactions must be confirmed across all databases before completion, or asynchronous, where transactions are replicated after being committed in the primary database.

The choice between synchronous and asynchronous replication depends on the specific requirements of data consistency and system performance.Data versioning is also employed to manage changes over time, allowing systems to track and manage updates to data sets.

This approach helps in resolving conflicts in distributed environments by maintaining a history of changes.

Lastly, regular synchronization checks are essential to ensure that data remains consistent across all nodes in the distributed system.

These checks identify and resolve discrepancies, maintaining the overall integrity and consistency of the distributed database system.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

CodeOracle SQL

SQL

SELECT * FROM employees WHERE ROWNUM <= 5;

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

Data state before query executes.

Output / Afterresult

After

EMP_ID	EMP_NAME	SALARY
101	Alice Johnson	50,000
102	Bob Smith	62,000
103	Carol White	55,000

Top 3 rows — ROWNUM/FETCH FIRST limits result set

Key Takeaway

Ensuring data consistency across distributed databases in SQL involves key techniques are distributed transactions, replication, conflict resolution and data versioning.

Big Data, NoSQL & Modern Data Architectures

Q117. What are the considerations for choosing a SQL vs. NoSQL database?

Answer

The choice between SQL and NoSQL databases hinges on your specific project requirements.

SQL databases are ideal for structured data, complex queries, and data integrity, while NoSQL databases offer flexibility, scalability, and adaptability to changing data needs.

Careful evaluation of your project's characteristics and goals will guide you to the most suitable database solution.SQL databases, known for their structured and relational data storage, excel in scenarios where data integrity, consistency, and complex queries are paramount.

They are well-suited for applications that demand ACID (Atomicity, Consistency, Isolation, Durability) compliance, such as financial systems and traditional relational data models.

SQL databases offer a robust schema that ensures data conformity and enforces relationships between tables.NoSQL databases provide greater flexibility and scalability, making them ideal for scenarios where rapid growth and diverse data types are anticipated.

They shine in applications requiring high throughput and low-latency data access, such as social media platforms and IoT (Internet of Things) applications.

NoSQL databases handle semi-structured and unstructured data effectively, making them suitable for dynamic data environments.Another crucial consideration is the need for horizontal scalability.

SQL databases typically scale vertically by adding more powerful hardware, while NoSQL databases excel at horizontal scaling by distributing data across multiple servers or nodes, enabling seamless expansion as data volume increases.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

CodeOracle SQL

SQL

SELECT * FROM employees WHERE ROWNUM <= 5;

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

Data state before query executes.

Output / Afterresult

After

EMP_ID	EMP_NAME	SALARY
101	Alice Johnson	50,000
102	Bob Smith	62,000
103	Carol White	55,000

Top 3 rows — ROWNUM/FETCH FIRST limits result set

Key Takeaway

The choice between SQL and NoSQL databases hinges on your specific project requirements.

Q118. What are the emerging trends and technologies in SQL database management?

JSON, XML & Semi-Structured Data

Q119. Explain the use of XML and JSON data types in SQL.

Answer

SQL's XML and JSON data types are essential tools for managing structured data within relational databases.

They enable the storage and manipulation of XML and JSON data, streamlining data integration and enhancing the capabilities of SQL databases in handling diverse data formats.XML data types allow SQL databases to store data in the Extensible Markup Language (XML) format.

XML is a hierarchical and self-descriptive format that is particularly useful for representing complex data structures.

SQL developers store, by using XML data types, query, and manipulate XML data directly within the database.

This capability is especially beneficial when dealing with data from web services or applications that communicate using XML.JSON data types enable SQL databases to work with data in JavaScript Object Notation (JSON) format.

JSON is a lightweight and flexible format that is widely used for data interchange in web applications.

SQL's support for JSON data types allows for the storage and querying of JSON documents, making it easier to work with data generated by web APIs or NoSQL databases.The use of XML and JSON data types in SQL provides several advantages.

It allows for better integration of data from diverse sources, as XML and JSON are common formats for data exchange on the web.

It also simplifies data manipulation and querying, enabling developers to extract specific information from complex documents efficiently.

It enhances the versatility of SQL databases, making them more adaptable to modern data-driven applications.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

CodeOracle SQL

SQL

SELECT * FROM employees WHERE ROWNUM <= 5;

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

Data state before query executes.

Output / Afterresult

After

EMP_ID	EMP_NAME	SALARY
101	Alice Johnson	50,000
102	Bob Smith	62,000
103	Carol White	55,000

Top 3 rows — ROWNUM/FETCH FIRST limits result set

Key Takeaway

SQL's XML and JSON data types are essential tools for managing structured data within relational databases.

Q120. What are non-relational features in SQL, like JSON support?

Q121. How do you work with JSON data in Oracle Database?

Answer

Oracle provides JSON functions and operators to store, query, and manipulate JSON data efficiently using SQL.

CodeOracle SQL

SQL

-- Create table with JSON column
CREATE TABLE customer_data (
   customer_id NUMBER,
   customer_info JSON
);

-- Insert JSON data
INSERT INTO customer_data VALUES (
   1, 
   JSON_OBJECT(
       'name' VALUE 'John Doe',
       'age' VALUE 30,
       'email' VALUE 'john@email.com',
       'address' VALUE JSON_OBJECT(
           'street' VALUE '123 Main St',
           'city' VALUE 'New York',
           'zipcode' VALUE '10001'
       ),
       'orders' VALUE JSON_ARRAY(101, 102, 103)
   )
);

-- Query JSON data using dot notation
SELECT customer_id,
      customer_info.name as customer_name,
      customer_info.age as age,
      customer_info.address.city as city
FROM customer_data;

-- Query JSON arrays
SELECT customer_id,
      customer_info.name as customer_name,
      jo.order_id
FROM customer_data c,
    JSON_TABLE(c.customer_info, '$.orders[*]' 
               COLUMNS (order_id NUMBER PATH '$')) jo;

-- Complex JSON queries with JSON_TABLE
SELECT customer_id, name, age, street, city, order_id
FROM customer_data,
    JSON_TABLE(customer_info, '$'
               COLUMNS (
                   name VARCHAR2(100) PATH '$.name',
                   age NUMBER PATH '$.age',
                   street VARCHAR2(200) PATH '$.address.street',
                   city VARCHAR2(100) PATH '$.address.city',
                   NESTED PATH '$.orders[*]' COLUMNS (
                       order_id NUMBER PATH '$'
                   )
               ));

-- JSON aggregation
SELECT JSON_OBJECT(
   'total_customers' VALUE COUNT(*),
   'avg_age' VALUE AVG(customer_info.age.number()),
   'cities' VALUE JSON_ARRAYAGG(DISTINCT customer_info.address.city.string())
) as summary
FROM customer_data;

-- Update JSON data
UPDATE customer_data
SET customer_info = JSON_MERGEPATCH(
   customer_info,
   JSON_OBJECT('phone' VALUE '555-1234')
)
WHERE customer_id = 1;

-- Search within JSON
SELECT customer_id, customer_info.name
FROM customer_data
WHERE JSON_EXISTS(customer_info, '$.address.city?(@ == "New York")');

-- Create indexes on JSON data
CREATE INDEX idx_customer_name ON customer_data (JSON_VALUE(customer_info, '$.name'));
CREATE INDEX idx_customer_city ON customer_data (JSON_VALUE(customer_info, '$.address.city'));

Output / Afterresult

After

COUNT	AVG_SALARY	MIN_SALARY	MAX_SALARY	SUM_SALARY
4	59,250	50,000	70,000	237,000

Aggregate over all 4 employees — AVG=(50+62+55+70)/4=59,250

Key Takeaway

Oracle provides JSON functions and operators to store, query, and manipulate JSON data efficiently using SQL.

Backup, Recovery & Auditing

Q122. How do you handle large-scale data migrations in SQL?

Answer

Handling large-scale data migrations in SQL requires careful planning and execution to ensure the integrity of the data and the efficiency of the process.

One fundamental approach is to leverage SQL's built-in features and best practices.It's crucial to utilize SQL's transactional capabilities.

Transactions allow you to group multiple SQL statements into a single unit of work.

Ensure that either all changes are applied successfully or none at all, maintaining data consistency, by wrapping your data migration operations in transactions.SQL provides powerful tools for data transformation.

SQL's SELECT INTO and INSERT INTO statements allow you to extract and load data efficiently between tables.

SQL's JOIN operations enable you to combine data from multiple sources into a single coherent dataset, facilitating complex migrations.Consider using ETL (Extract, Transform, Load) tools that are designed specifically for large-scale data migrations.

These tools streamline the process by automating data extraction, transformation, and loading tasks, reducing the risk of errors and optimizing performance.Handling large-scale data migrations in SQL involves leveraging SQL's transactional capabilities, utilizing data transformation operations, and considering specialized ETL tools.

This approach ensures a smooth and reliable migration process while maintaining data integrity and performance.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

CodeOracle SQL

SQL

SELECT * FROM employees WHERE ROWNUM <= 5;

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

Data state before query executes.

Output / Afterresult

After

EMP_ID	EMP_NAME	SALARY
101	Alice Johnson	50,000
102	Bob Smith	62,000
103	Carol White	55,000

Top 3 rows — ROWNUM/FETCH FIRST limits result set

Key Takeaway

Handling large-scale data migrations in SQL requires careful planning and execution to ensure the integrity of the data and the efficiency of the process.

Q123. How do you manage backup and recovery in SQL databases?

Answer

Managing backup and recovery in SQL databases involves several key strategies to ensure data integrity and availability.

The core approach includes regularly scheduled backups, which consist of full backups, differential backups, and transaction log backups.

Full backups capture the entire database at a given point in time, providing a comprehensive data snapshot.

Differential backups record changes made since the last full backup, offering a balance between storage efficiency and recovery speed.

Transaction log backups, essential for databases in full recovery mode, capture every transaction in the database, allowing for point-in-time recovery and minimal data loss in case of a failure.Recovery in SQL databases utilizes these backups to restore the database to a specific state.

The recovery process typically starts with the most recent full backup, In the event of data loss or corruption, followed by the application of the latest differential backup and then the relevant transaction log backups up to the point of failure or a specific point in time.

This method ensures data recovery with minimal loss.

The effectiveness of the backup and recovery process depends on regular testing and validation of backups, ensuring they are complete and restorable.

Automating the backup process and monitoring backup health are also crucial steps in maintaining a robust backup and recovery strategy for SQL databases.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

CodeOracle SQL

SQL

SELECT * FROM employees WHERE ROWNUM <= 5;

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

Data state before query executes.

Output / Afterresult

After

EMP_ID	EMP_NAME	SALARY
101	Alice Johnson	50,000
102	Bob Smith	62,000
103	Carol White	55,000

Top 3 rows — ROWNUM/FETCH FIRST limits result set

Key Takeaway

Managing backup and recovery in SQL databases involves several key strategies to ensure data integrity and availability.

Q124. Discuss the importance and methods of database auditing.

Q125. Describe strategies for efficient data archiving in SQL databases.

Q126. How can you backup and restore an Oracle database?

Answer

Oracle provides multiple backup and recovery methods including RMAN (Recovery Manager), Data Pump, and traditional export/import utilities.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

CodeOracle SQL

SQL

-- 1. RMAN (Recovery Manager) - Recommended method
-- Connect to RMAN
-- rman target /

-- Full database backup
BACKUP DATABASE;

-- Incremental backup
BACKUP INCREMENTAL LEVEL 1 DATABASE;

-- Backup specific tablespace
BACKUP TABLESPACE users;

-- Backup archive logs
BACKUP ARCHIVELOG ALL;

-- 2. Data Pump Export/Import
-- Full database export (as SYSDBA)
-- expdp system/password FULL=Y DIRECTORY=DATA_PUMP_DIR DUMPFILE=fulldb.dmp

-- Schema export
-- expdp hr/password SCHEMAS=hr DIRECTORY=DATA_PUMP_DIR DUMPFILE=hr_schema.dmp

-- Table export
-- expdp hr/password TABLES=employees,departments DIRECTORY=DATA_PUMP_DIR DUMPFILE=hr_tables.dmp

-- Import examples
-- impdp system/password FULL=Y DIRECTORY=DATA_PUMP_DIR DUMPFILE=fulldb.dmp

-- Schema import
-- impdp system/password SCHEMAS=hr DIRECTORY=DATA_PUMP_DIR DUMPFILE=hr_schema.dmp

-- 3. SQL*Plus Export (deprecated but still used)
-- Create directory object
CREATE OR REPLACE DIRECTORY backup_dir AS '/u01/backup';

-- Traditional export/import (exp/imp utilities)
-- exp hr/password file=hr_backup.dmp owner=hr
-- imp hr/password file=hr_backup.dmp fromuser=hr touser=hr_new

-- 4. Logical backup using SQL*Plus
SPOOL /u01/backup/employees_backup.sql
SELECT 'INSERT INTO employees VALUES (' ||
       employee_id || ',' ||
       '''' || first_name || ''',' ||
       '''' || last_name || ''',' ||
       '''' || email || ''',' ||
       'DATE ''' || TO_CHAR(hire_date, 'YYYY-MM-DD') || ''');'
FROM employees;
SPOOL OFF;

-- 5. Flashback Database (Oracle 10g+)
-- Enable flashback database
ALTER DATABASE FLASHBACK ON;

-- Create restore point
CREATE RESTORE POINT before_maintenance;

-- Flashback database to restore point
FLASHBACK DATABASE TO RESTORE POINT before_maintenance;

-- Flashback table to specific time
FLASHBACK TABLE employees TO TIMESTAMP 
(SYSTIMESTAMP - INTERVAL '1' HOUR);

-- 6. RMAN Recovery scenarios
-- Recover database
RECOVER DATABASE;

-- Point-in-time recovery
RECOVER DATABASE UNTIL TIME '2024-01-15 14:30:00';

-- Recover specific tablespace
RECOVER TABLESPACE users;

-- 7. Cold backup (database shutdown)
-- Shutdown database
SHUTDOWN IMMEDIATE;

-- Copy all datafiles, control files, and redo logs
-- cp /u01/app/oracle/oradata/ORCL/*.dbf /backup/
-- cp /u01/app/oracle/oradata/ORCL/*.ctl /backup/
-- cp /u01/app/oracle/oradata/ORCL/*.log /backup/

-- Startup database
STARTUP;

-- 8. Check backup status
SELECT session_key, input_type, status, 
       start_time, end_time, 
       input_bytes/1024/1024 as input_mb
FROM v$rman_backup_job_details
ORDER BY start_time DESC;

-- 9. Validate backups
VALIDATE BACKUPSET ALL;

-- Cross-check backups
CROSSCHECK BACKUP;

-- 10. Archive log management
-- Enable archive log mode
SHUTDOWN IMMEDIATE;
STARTUP MOUNT;
ALTER DATABASE ARCHIVELOG;
ALTER DATABASE OPEN;

-- Check archive log status
SELECT log_mode FROM v$database;

-- Backup and delete archive logs
BACKUP ARCHIVELOG ALL DELETE INPUT;

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

Data state before query executes.

Output / Afterresult

After

EMP_ID	EMP_NAME	DEPT_ID	SALARY
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	20	70,000

All 4 rows returned — ordered by salary DESC

Key Takeaway

Oracle provides multiple backup and recovery methods including RMAN (Recovery Manager), Data Pump, and traditional export/import utilities.

Q127. What is Oracle Flashback Query and how is it used for recovery?

Oracle-Specific Practical Queries

Q128. How can you find the Nth highest salary from a table in Oracle?

Q129. What is the difference between CHAR and VARCHAR2 data types in Oracle?

Q130. Write a query to retrieve duplicate records from a table without using the DISTINCT keyword.

Q131. How can you fetch alternate records from a table in Oracle?

Q132. Write a query to find the second highest salary from a table in Oracle.

Answer

Use ranking functions or subqueries to find the second highest salary, handling potential duplicates appropriately.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

CodeOracle SQL

SQL

-- Method 1: Using ROW_NUMBER() (most common)
SELECT salary 
FROM (
    SELECT salary, ROW_NUMBER() OVER (ORDER BY salary DESC) as rn
    FROM employee
    WHERE salary IS NOT NULL
)
WHERE rn = 2;

-- Method 2: Using DENSE_RANK() (handles duplicates better)
SELECT DISTINCT salary 
FROM (
    SELECT salary, DENSE_RANK() OVER (ORDER BY salary DESC) as rnk
    FROM employee
    WHERE salary IS NOT NULL
)
WHERE rnk = 2;

-- Method 3: Using MAX with subquery
SELECT MAX(salary) as second_highest_salary
FROM employee 
WHERE salary < (SELECT MAX(salary) FROM employee);

-- Method 4: Using ROWNUM (Oracle specific)
SELECT salary
FROM (
    SELECT DISTINCT salary
    FROM employee
    WHERE salary IS NOT NULL
    ORDER BY salary DESC
)
WHERE ROWNUM <= 2
MINUS
SELECT salary
FROM (
    SELECT DISTINCT salary
    FROM employee
    WHERE salary IS NOT NULL
    ORDER BY salary DESC
)
WHERE ROWNUM <= 1;

-- Method 5: Using OFFSET FETCH (Oracle 12c+)
SELECT salary
FROM (
    SELECT DISTINCT salary
    FROM employee
    WHERE salary IS NOT NULL
    ORDER BY salary DESC
)
OFFSET 1 ROWS FETCH NEXT 1 ROWS ONLY;

-- Method 6: With employee details
SELECT employee_id, first_name, last_name, salary
FROM (
    SELECT employee_id, first_name, last_name, salary,
           RANK() OVER (ORDER BY salary DESC) as salary_rank
    FROM employee
    WHERE salary IS NOT NULL
)
WHERE salary_rank = 2;

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

Data state before query executes.

Output / Afterresult

After

EMP_NAME	SALARY	ROW_NUM	RANK	DENSE_RANK
Dave Brown	70,000	1	1	1
Bob Smith	62,000	2	2	2
Carol White	55,000	3	3	3
Alice Johnson	50,000	4	4	4

All 4 employees ranked by salary DESC — no ties so all methods agree

Key Takeaway

Use ranking functions or subqueries to find the second highest salary, handling potential duplicates appropriately.

Q133. Write a query to find all employees who started after Jan 1, 2020, but before Jan 1, 2023 in Oracle.

Answer

Use date comparison with proper Oracle date formatting and BETWEEN clause or comparison operators.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

CodeOracle SQL

SQL

-- Method 1: Using BETWEEN clause
SELECT employee_id, first_name, last_name, hire_date
FROM employees 
WHERE hire_date BETWEEN DATE '2020-01-01' AND DATE '2022-12-31';

-- Method 2: Using comparison operators
SELECT employee_id, first_name, last_name, hire_date
FROM employees 
WHERE hire_date >= DATE '2020-01-01' 
  AND hire_date < DATE '2023-01-01';

-- Method 3: Using TO_DATE function
SELECT employee_id, first_name, last_name, hire_date
FROM employees 
WHERE hire_date >= TO_DATE('01-JAN-2020', 'DD-MON-YYYY')
  AND hire_date < TO_DATE('01-JAN-2023', 'DD-MON-YYYY');

-- Method 4: With formatted output
SELECT employee_id, 
       first_name, 
       last_name, 
       TO_CHAR(hire_date, 'DD-MON-YYYY') as formatted_hire_date,
       ROUND(MONTHS_BETWEEN(SYSDATE, hire_date)/12, 1) as years_of_service
FROM employees 
WHERE hire_date BETWEEN DATE '2020-01-01' AND DATE '2022-12-31'
ORDER BY hire_date;

-- Method 5: Using EXTRACT function
SELECT employee_id, first_name, last_name, hire_date
FROM employees 
WHERE EXTRACT(YEAR FROM hire_date) BETWEEN 2020 AND 2022
  AND NOT (EXTRACT(YEAR FROM hire_date) = 2023 AND EXTRACT(MONTH FROM hire_date) = 1 AND EXTRACT(DAY FROM hire_date) = 1);

-- Method 6: With additional date calculations
SELECT employee_id, 
       first_name, 
       last_name, 
       hire_date,
       CASE 
           WHEN EXTRACT(YEAR FROM hire_date) = 2020 THEN 'Hired in 2020'
           WHEN EXTRACT(YEAR FROM hire_date) = 2021 THEN 'Hired in 2021'
           WHEN EXTRACT(YEAR FROM hire_date) = 2022 THEN 'Hired in 2022'
       END as hire_year_group
FROM employees 
WHERE hire_date >= DATE '2020-01-01' 
  AND hire_date < DATE '2023-01-01'
ORDER BY hire_date;

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

Data state before query executes.

Output / Afterresult

After

EMP_ID	EMP_NAME	DEPT_ID	SALARY
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	20	70,000

All 4 rows returned — ordered by salary ASC

Key Takeaway

Use date comparison with proper Oracle date formatting and BETWEEN clause or comparison operators.

Q134. How can you implement pagination in Oracle SQL?

Answer

Oracle provides multiple methods for pagination including ROWNUM, ROW_NUMBER(), and the modern OFFSET FETCH syntax (Oracle 12c+).

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

CodeOracle SQL

SQL

-- Method 1: Using ROWNUM (Oracle 11g and earlier)
-- Page 1 (records 1-10)
SELECT * FROM (
    SELECT e.*, ROWNUM rn FROM (
        SELECT employee_id, first_name, last_name, salary
        FROM employees
        ORDER BY salary DESC
    ) e
    WHERE ROWNUM <= 10
) WHERE rn >= 1;

-- Page 2 (records 11-20)
SELECT * FROM (
    SELECT e.*, ROWNUM rn FROM (
        SELECT employee_id, first_name, last_name, salary
        FROM employees
        ORDER BY salary DESC
    ) e
    WHERE ROWNUM <= 20
) WHERE rn >= 11;

-- Method 2: Using ROW_NUMBER() Analytical Function
-- Page 1 (records 1-10)
SELECT * FROM (
    SELECT employee_id, first_name, last_name, salary,
           ROW_NUMBER() OVER (ORDER BY salary DESC) as rn
    FROM employees
)
WHERE rn BETWEEN 1 AND 10;

-- Page 2 (records 11-20)
SELECT * FROM (
    SELECT employee_id, first_name, last_name, salary,
           ROW_NUMBER() OVER (ORDER BY salary DESC) as rn
    FROM employees
)
WHERE rn BETWEEN 11 AND 20;

-- Method 3: OFFSET FETCH (Oracle 12c+) - Recommended
-- Page 1 (first 10 records)
SELECT employee_id, first_name, last_name, salary
FROM employees
ORDER BY salary DESC
OFFSET 0 ROWS FETCH NEXT 10 ROWS ONLY;

-- Page 2 (next 10 records)
SELECT employee_id, first_name, last_name, salary
FROM employees
ORDER BY salary DESC
OFFSET 10 ROWS FETCH NEXT 10 ROWS ONLY;

-- Page 3 (next 10 records)
SELECT employee_id, first_name, last_name, salary
FROM employees
ORDER BY salary DESC
OFFSET 20 ROWS FETCH NEXT 10 ROWS ONLY;

-- Parameterized Pagination Function
CREATE OR REPLACE FUNCTION get_employees_page(
    p_page_number NUMBER,
    p_page_size NUMBER DEFAULT 10
) RETURN SYS_REFCURSOR IS
    v_cursor SYS_REFCURSOR;
    v_offset NUMBER;
BEGIN
    v_offset := (p_page_number - 1) * p_page_size;
    
    OPEN v_cursor FOR
        SELECT employee_id, first_name, last_name, salary
        FROM employees
        ORDER BY employee_id
        OFFSET v_offset ROWS FETCH NEXT p_page_size ROWS ONLY;
    
    RETURN v_cursor;
END;
/

-- Pagination with Total Count
SELECT * FROM (
    SELECT employee_id, first_name, last_name, salary,
           ROW_NUMBER() OVER (ORDER BY salary DESC) as rn,
           COUNT(*) OVER () as total_count
    FROM employees
)
WHERE rn BETWEEN 11 AND 20;

-- Dynamic Pagination PL/SQL Block
DECLARE
    v_page_number NUMBER := 2;
    v_page_size NUMBER := 5;
    v_offset NUMBER;
    v_sql VARCHAR2(1000);
    TYPE emp_cursor_type IS REF CURSOR;
    emp_cursor emp_cursor_type;
    v_emp_id NUMBER;
    v_name VARCHAR2(100);
    v_salary NUMBER;
BEGIN
    v_offset := (v_page_number - 1) * v_page_size;
    
    v_sql := 'SELECT employee_id, first_name || '' '' || last_name, salary
              FROM employees
              ORDER BY salary DESC
              OFFSET :1 ROWS FETCH NEXT :2 ROWS ONLY';
    
    OPEN emp_cursor FOR v_sql USING v_offset, v_page_size;
    
    DBMS_OUTPUT.PUT_LINE('Page ' || v_page_number || ' (Page Size: ' || v_page_size || ')');
    DBMS_OUTPUT.PUT_LINE('----------------------------------------');
    
    LOOP
        FETCH emp_cursor INTO v_emp_id, v_name, v_salary;
        EXIT WHEN emp_cursor%NOTFOUND;
        
        DBMS_OUTPUT.PUT_LINE(v_emp_id || ' - ' || v_name || ' -  || v_salary);
    END LOOP;
    
    CLOSE emp_cursor;
END;
/

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

Data state before query executes.

Output / Afterresult

After

EMP_NAME	SALARY	ROW_NUM	RANK	DENSE_RANK
Dave Brown	70,000	1	1	1
Bob Smith	62,000	2	2	2
Carol White	55,000	3	3	3
Alice Johnson	50,000	4	4	4

All 4 employees ranked by salary DESC — no ties so all methods agree

Key Takeaway

Oracle provides multiple methods for pagination including ROWNUM, ROW_NUMBER(), and the modern OFFSET FETCH syntax (Oracle 12c+).

Q135. How can you find all tables that have specific column names in an Oracle database?

Answer

Use Oracle data dictionary views like ALL_TAB_COLUMNS or USER_TAB_COLUMNS to search for tables containing specific column names.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

CodeOracle SQL

SQL

-- Find tables with specific column name
SELECT table_name, owner
FROM all_tab_columns
WHERE column_name = 'EMPLOYEE_ID'
ORDER BY owner, table_name;

-- Find tables in current schema with specific column
SELECT table_name, column_name, data_type, data_length
FROM user_tab_columns
WHERE column_name = 'SALARY'
ORDER BY table_name;

-- Find tables with multiple specific columns
SELECT table_name
FROM user_tab_columns
WHERE column_name IN ('FIRST_NAME', 'LAST_NAME')
GROUP BY table_name
HAVING COUNT(DISTINCT column_name) = 2;

-- Find tables with column name pattern
SELECT table_name, column_name
FROM user_tab_columns
WHERE column_name LIKE '%_DATE'
ORDER BY table_name, column_name;

-- Find tables with columns containing specific text
SELECT table_name, column_name, data_type
FROM user_tab_columns
WHERE column_name LIKE '%EMAIL%'
   OR column_name LIKE '%PHONE%'
ORDER BY table_name, column_name;

-- More detailed column information
SELECT table_name,
       column_name,
       data_type,
       CASE 
           WHEN data_type = 'VARCHAR2' THEN data_type || '(' || data_length || ')'
           WHEN data_type = 'NUMBER' THEN 
               CASE 
                   WHEN data_scale IS NULL THEN data_type || '(' || data_precision || ')'
                   ELSE data_type || '(' || data_precision || ',' || data_scale || ')'
               END
           ELSE data_type
       END as full_data_type,
       nullable,
       data_default
FROM user_tab_columns
WHERE column_name = 'DEPARTMENT_ID'
ORDER BY table_name;

-- Find foreign key relationships
SELECT a.table_name as child_table,
       a.column_name as child_column,
       a.constraint_name,
       c_pk.table_name as parent_table,
       c_pk.column_name as parent_column
FROM user_cons_columns a
JOIN user_constraints c ON a.owner = c.owner AND a.constraint_name = c.constraint_name
JOIN user_constraints c_pk ON c.r_owner = c_pk.owner AND c.r_constraint_name = c_pk.constraint_name
JOIN user_cons_columns c_pk ON c_pk.owner = c_pk.owner AND c_pk.constraint_name = c_pk.constraint_name
WHERE c.constraint_type = 'R'
  AND a.column_name = 'DEPARTMENT_ID'
ORDER BY a.table_name;

-- Find all columns for a specific table
SELECT column_id,
       column_name,
       data_type,
       CASE 
           WHEN data_type = 'VARCHAR2' THEN data_type || '(' || data_length || ')'
           WHEN data_type = 'NUMBER' THEN 
               CASE 
                   WHEN data_scale IS NULL THEN data_type
                   WHEN data_scale = 0 THEN data_type || '(' || data_precision || ')'
                   ELSE data_type || '(' || data_precision || ',' || data_scale || ')'
               END
           ELSE data_type
       END as formatted_type,
       nullable,
       data_default
FROM user_tab_columns
WHERE table_name = 'EMPLOYEES'
ORDER BY column_id;

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

Data state before query executes.

Output / Afterresult

After

EMP_NAME	DEPT_NAME
Alice Johnson	Engineering
Bob Smith	Marketing
Carol White	Engineering

3 rows — only rows with matching dept_id in both tables (Dave Brown excluded — no dept)

Key Takeaway

Use Oracle data dictionary views like ALL_TAB_COLUMNS or USER_TAB_COLUMNS to search for tables containing specific column names.

Q136. How do you concatenate strings in Oracle SQL?

Answer

Oracle provides multiple methods for string concatenation including the || operator, CONCAT function, and LISTAGG for multiple values.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

CodeOracle SQL

SQL

-- Method 1: Using || operator (most common)
SELECT first_name || ' ' || last_name AS full_name 
FROM employees;

-- Method 2: Using CONCAT function (only 2 arguments)
SELECT CONCAT(first_name, last_name) AS full_name
FROM employees;

-- Method 3: Nested CONCAT for multiple strings
SELECT CONCAT(CONCAT(first_name, ' '), last_name) AS full_name
FROM employees;

-- Method 4: Combining || with other functions
SELECT first_name || ' ' || last_name || ' (ID: ' || employee_id || ')' AS employee_info
FROM employees;

-- Method 5: Handling NULL values
SELECT first_name || ' ' || NVL(middle_name || ' ', '') || last_name AS full_name
FROM employees;

-- Method 6: Using LISTAGG for multiple rows
SELECT department_id,
       LISTAGG(first_name, ', ') WITHIN GROUP (ORDER BY first_name) AS employee_names
FROM employees
GROUP BY department_id;

-- Method 7: Advanced concatenation with formatting
SELECT 'Employee: ' || INITCAP(first_name) || ' ' || UPPER(last_name) || 
       ' earns  || TO_CHAR(salary, '999,999.99') || ' per year' AS employee_summary
FROM employees;

-- Method 8: Conditional concatenation
SELECT first_name || 
       CASE 
           WHEN middle_name IS NOT NULL THEN ' ' || middle_name || ' '
           ELSE ' '
       END || last_name AS full_name
FROM employees;

-- Method 9: XML concatenation for complex scenarios
SELECT employee_id,
       XMLAGG(XMLELEMENT("skill", skill_name || '; ')).getClobVal() AS skills
FROM employee_skills
GROUP BY employee_id;

-- Method 10: Using REPLACE for clean concatenation
SELECT REPLACE(first_name || ' ' || middle_name || ' ' || last_name, '  ', ' ') AS clean_name
FROM employees;

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

Data state before query executes.

Output / Afterresult

After

DEPT_ID	EMPLOYEE_NAMES
10	Alice Johnson, Carol White
20	Bob Smith, Dave Brown

LISTAGG aggregates names into comma-separated string per group

Key Takeaway

Oracle provides multiple methods for string concatenation including the || operator, CONCAT function, and LISTAGG for multiple values.

Q137. How can you retrieve unique values from a column without using the DISTINCT keyword in Oracle?

Answer

Use GROUP BY clause to retrieve unique values from a column, which can also provide additional aggregation information.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

CodeOracle SQL

SQL

-- Method 1: Using GROUP BY
SELECT department_id 
FROM employees 
GROUP BY department_id
ORDER BY department_id;

-- Method 2: GROUP BY with COUNT
SELECT department_id, COUNT(*) as employee_count
FROM employees 
GROUP BY department_id
ORDER BY department_id;

-- Method 3: Using UNION (removes duplicates by default)
SELECT department_id FROM employees WHERE department_id <= 50
UNION
SELECT department_id FROM employees WHERE department_id > 50;

-- Method 4: Using analytical functions with ROW_NUMBER
SELECT department_id
FROM (
    SELECT department_id,
           ROW_NUMBER() OVER (PARTITION BY department_id ORDER BY department_id) as rn
    FROM employees
)
WHERE rn = 1;

-- Method 5: Using EXISTS subquery
SELECT DISTINCT e1.department_id
FROM employees e1
WHERE EXISTS (
    SELECT 1 FROM employees e2 
    WHERE e2.department_id = e1.department_id
);

-- Method 6: Using IN with subquery
SELECT department_id
FROM employees
WHERE department_id IN (SELECT department_id FROM employees)
GROUP BY department_id;

-- Method 7: Using MINUS to find unique values
SELECT department_id FROM employees
MINUS
SELECT department_id FROM employees e1
WHERE EXISTS (
    SELECT 1 FROM employees e2 
    WHERE e2.department_id = e1.department_id 
    AND e2.ROWID < e1.ROWID
);

-- Method 8: Using LISTAGG for unique concatenated values
SELECT LISTAGG(DISTINCT department_id, ',') WITHIN GROUP (ORDER BY department_id) as unique_departments
FROM employees;

-- Method 9: Using COLLECT for unique values as array
SELECT CAST(COLLECT(DISTINCT department_id) AS sys.odcinumberlist) as unique_dept_list
FROM employees;

-- Method 10: Multiple column uniqueness using GROUP BY
SELECT first_name, last_name, COUNT(*) as occurrence_count
FROM employees
GROUP BY first_name, last_name
ORDER BY first_name, last_name;

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

Data state before query executes.

Output / Afterresult

After

EMP_NAME	SALARY	ROW_NUM	RANK	DENSE_RANK
Dave Brown	70,000	1	1	1
Bob Smith	62,000	2	2	2
Carol White	55,000	3	3	3
Alice Johnson	50,000	4	4	4

All 4 employees ranked by salary DESC — no ties so all methods agree

Key Takeaway

Use GROUP BY clause to retrieve unique values from a column, which can also provide additional aggregation information.

Q138. How can you create a copy of a table, including both structure and data, in Oracle?

Answer

Use CREATE TABLE AS SELECT (CTAS) statement to create a copy of a table with both structure and data in Oracle.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

CodeOracle SQL

SQL

-- Method 1: Create complete copy with data
CREATE TABLE employees_copy AS SELECT * FROM employees;

-- Method 2: Create copy with specific columns
CREATE TABLE employees_basic AS 
SELECT employee_id, first_name, last_name, salary 
FROM employees;

-- Method 3: Create copy with filtered data
CREATE TABLE high_salary_employees AS 
SELECT * FROM employees 
WHERE salary > 50000;

-- Method 4: Create copy with additional computed columns
CREATE TABLE employees_enhanced AS 
SELECT employee_id,
       first_name,
       last_name,
       salary,
       salary * 12 as annual_salary,
       CASE 
           WHEN salary > 75000 THEN 'High'
           WHEN salary > 50000 THEN 'Medium'
           ELSE 'Low'
       END as salary_grade
FROM employees;

-- Method 5: Create empty copy (structure only)
CREATE TABLE employees_structure AS 
SELECT * FROM employees WHERE 1=0;

-- Method 6: Create copy with joins
CREATE TABLE employee_department_copy AS
SELECT e.employee_id,
       e.first_name,
       e.last_name,
       e.salary,
       d.department_name,
       d.location_id
FROM employees e
JOIN departments d ON e.department_id = d.department_id;

-- Method 7: Create copy with specific tablespace
CREATE TABLE employees_copy_ts
TABLESPACE users
AS SELECT * FROM employees;

-- Method 8: Create copy with storage parameters
CREATE TABLE employees_copy_storage (
    INITIAL 1M
    NEXT 1M
    PCTINCREASE 0
)
AS SELECT * FROM employees;

-- Note: CTAS limitations
-- 1. Does not copy indexes
-- 2. Does not copy constraints (except NOT NULL)
-- 3. Does not copy triggers
-- 4. Does not copy comments

-- Copy indexes separately
CREATE INDEX idx_emp_copy_name ON employees_copy(last_name);

-- Copy constraints separately
ALTER TABLE employees_copy ADD CONSTRAINT pk_emp_copy PRIMARY KEY (employee_id);

-- Copy comments separately
COMMENT ON TABLE employees_copy IS 'Copy of employees table';
COMMENT ON COLUMN employees_copy.employee_id IS 'Unique employee identifier';

-- Alternative: Use DBMS_METADATA for complete structure
-- Generate DDL for original table
SELECT DBMS_METADATA.GET_DDL('TABLE', 'EMPLOYEES') FROM dual;

-- Method 9: Create copy with specific conditions and ordering
CREATE TABLE recent_employees AS
SELECT * FROM employees
WHERE hire_date >= ADD_MONTHS(SYSDATE, -24)
ORDER BY hire_date DESC;

-- Method 10: Create partitioned copy
CREATE TABLE employees_partitioned_copy
PARTITION BY RANGE (hire_date) (
    PARTITION p_old VALUES LESS THAN (DATE '2020-01-01'),
    PARTITION p_recent VALUES LESS THAN (DATE '2025-01-01')
)
AS SELECT * FROM employees;

-- Verify copy
SELECT COUNT(*) FROM employees;
SELECT COUNT(*) FROM employees_copy;

-- Compare structures
DESC employees;
DESC employees_copy;

-- Check if data matches
SELECT COUNT(*) FROM employees
MINUS
SELECT COUNT(*) FROM employees_copy;

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

Data state before query executes.

Output / Afterresult

After

EMP_NAME	DEPT_NAME
Alice Johnson	Engineering
Bob Smith	Marketing
Carol White	Engineering

3 rows — only rows with matching dept_id in both tables (Dave Brown excluded — no dept)

Key Takeaway

Use CREATE TABLE AS SELECT (CTAS) statement to create a copy of a table with both structure and data in Oracle.

Q139. How can you convert a UNIX timestamp into a readable date format in Oracle?

Answer

Use FROM_UNIXTIME equivalent functions or date arithmetic to convert UNIX timestamps to Oracle DATE format.

CodeOracle SQL

SQL

-- Method 1: Convert UNIX timestamp to DATE
SELECT TO_DATE('1970-01-01', 'YYYY-MM-DD') + (1609459200 / 86400) as converted_date
FROM dual;

-- Method 2: Using TIMESTAMP conversion
SELECT TIMESTAMP '1970-01-01 00:00:00' + NUMTODSINTERVAL(1609459200, 'SECOND') as converted_timestamp
FROM dual;

-- Method 3: Create a reusable function
CREATE OR REPLACE FUNCTION unix_to_date(p_unix_timestamp NUMBER) 
RETURN DATE IS
BEGIN
    RETURN DATE '1970-01-01' + (p_unix_timestamp / 86400);
END;
/

-- Usage of the function
SELECT unix_to_date(1609459200) as readable_date FROM dual;

-- Method 4: Handle milliseconds UNIX timestamp
SELECT DATE '1970-01-01' + (1609459200000 / 86400000) as converted_date
FROM dual;

-- Method 5: Format the converted date
SELECT TO_CHAR(
    DATE '1970-01-01' + (1609459200 / 86400), 
    'YYYY-MM-DD HH24:MI:SS'
) as formatted_date
FROM dual;

-- Method 6: Convert multiple timestamps in a table
CREATE TABLE unix_timestamps (
    id NUMBER,
    unix_time NUMBER,
    description VARCHAR2(100)
);

INSERT INTO unix_timestamps VALUES (1, 1609459200, 'New Year 2021');
INSERT INTO unix_timestamps VALUES (2, 1640995200, 'New Year 2022');
INSERT INTO unix_timestamps VALUES (3, 1672531200, 'New Year 2023');

SELECT id,
       unix_time,
       DATE '1970-01-01' + (unix_time / 86400) as converted_date,
       TO_CHAR(DATE '1970-01-01' + (unix_time / 86400), 'DD-MON-YYYY HH24:MI:SS') as formatted_date,
       description
FROM unix_timestamps;

-- Method 7: Handle time zones
SELECT FROM_TZ(
    TIMESTAMP '1970-01-01 00:00:00' + NUMTODSINTERVAL(1609459200, 'SECOND'),
    'UTC'
) AT TIME ZONE 'America/New_York' as est_time
FROM dual;

-- Method 8: Reverse conversion (DATE to UNIX timestamp)
SELECT (SYSDATE - DATE '1970-01-01') * 86400 as unix_timestamp
FROM dual;

-- Method 9: Using EXTRACT for components
SELECT EXTRACT(YEAR FROM DATE '1970-01-01' + (1609459200 / 86400)) as year,
       EXTRACT(MONTH FROM DATE '1970-01-01' + (1609459200 / 86400)) as month,
       EXTRACT(DAY FROM DATE '1970-01-01' + (1609459200 / 86400)) as day
FROM dual;

-- Method 10: Create a comprehensive conversion function
CREATE OR REPLACE FUNCTION unix_to_formatted_date(
    p_unix_timestamp NUMBER,
    p_format VARCHAR2 DEFAULT 'YYYY-MM-DD HH24:MI:SS'
) RETURN VARCHAR2 IS
BEGIN
    RETURN TO_CHAR(
        DATE '1970-01-01' + (p_unix_timestamp / 86400),
        p_format
    );
END;
/

-- Usage examples
SELECT unix_to_formatted_date(1609459200) as default_format,
       unix_to_formatted_date(1609459200, 'DD-MON-YYYY') as custom_format,
       unix_to_formatted_date(1609459200, 'Day, DD Month YYYY') as verbose_format
FROM dual;

-- Clean up
DROP TABLE unix_timestamps;
DROP FUNCTION unix_to_date;
DROP FUNCTION unix_to_formatted_date;

Output / Afterresult

After

EMP_ID	EMP_NAME	DEPT_ID	SALARY
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	20	70,000

All 4 rows returned

Key Takeaway

Use FROM_UNIXTIME equivalent functions or date arithmetic to convert UNIX timestamps to Oracle DATE format.

Q140. What is the difference between SYSDATE and CURRENT_DATE in Oracle?

Answer

SYSDATE returns the current date and time of the database server, while CURRENT_DATE returns the current date and time in the session time zone.

CodeOracle SQL

SQL

-- SYSDATE: Database server date/time
SELECT SYSDATE as server_date_time FROM dual;

-- CURRENT_DATE: Session time zone date/time
SELECT CURRENT_DATE as session_date_time FROM dual;

-- CURRENT_TIMESTAMP: More precise session timestamp
SELECT CURRENT_TIMESTAMP as session_timestamp FROM dual;

-- SYSTIMESTAMP: Database server timestamp with time zone
SELECT SYSTIMESTAMP as server_timestamp FROM dual;

-- Show time zone information
SELECT SESSIONTIMEZONE as session_tz,
       DBTIMEZONE as database_tz
FROM dual;

-- Demonstrate difference with time zone change
ALTER SESSION SET TIME_ZONE = 'America/New_York';

SELECT SYSDATE as server_date,
       CURRENT_DATE as session_date,
       CURRENT_TIMESTAMP as session_timestamp
FROM dual;

-- Change to different time zone
ALTER SESSION SET TIME_ZONE = 'Asia/Tokyo';

SELECT SYSDATE as server_date,
       CURRENT_DATE as session_date,
       CURRENT_TIMESTAMP as session_timestamp
FROM dual;

-- Reset to default time zone
ALTER SESSION SET TIME_ZONE = DBTIMEZONE;

-- Date arithmetic examples
SELECT SYSDATE as today,
       SYSDATE + 1 as tomorrow,
       SYSDATE - 1 as yesterday,
       SYSDATE + 1/24 as one_hour_later,
       SYSDATE + 1/24/60 as one_minute_later
FROM dual;

-- Extract components
SELECT EXTRACT(YEAR FROM SYSDATE) as current_year,
       EXTRACT(MONTH FROM SYSDATE) as current_month,
       EXTRACT(DAY FROM SYSDATE) as current_day,
       EXTRACT(HOUR FROM CURRENT_TIMESTAMP) as current_hour,
       EXTRACT(MINUTE FROM CURRENT_TIMESTAMP) as current_minute
FROM dual;

-- Formatting dates
SELECT TO_CHAR(SYSDATE, 'YYYY-MM-DD HH24:MI:SS') as formatted_sysdate,
       TO_CHAR(CURRENT_DATE, 'YYYY-MM-DD HH24:MI:SS') as formatted_current_date,
       TO_CHAR(CURRENT_TIMESTAMP, 'YYYY-MM-DD HH24:MI:SS.FF3') as formatted_timestamp
FROM dual;

-- Practical usage in applications
CREATE TABLE event_log (
    event_id NUMBER,
    event_description VARCHAR2(200),
    server_time DATE DEFAULT SYSDATE,
    user_time DATE DEFAULT CURRENT_DATE,
    event_timestamp TIMESTAMP DEFAULT CURRENT_TIMESTAMP
);

INSERT INTO event_log (event_id, event_description) 
VALUES (1, 'User login event');

SELECT * FROM event_log;

-- Date comparisons
SELECT COUNT(*) as today_events
FROM event_log
WHERE TRUNC(server_time) = TRUNC(SYSDATE);

-- Time zone conversions
SELECT server_time,
       FROM_TZ(CAST(server_time AS TIMESTAMP), DBTIMEZONE) 
       AT TIME ZONE 'UTC' as utc_time,
       FROM_TZ(CAST(server_time AS TIMESTAMP), DBTIMEZONE) 
       AT TIME ZONE 'America/Los_Angeles' as pst_time
FROM event_log;

-- Performance consideration: SYSDATE vs CURRENT_DATE
-- SYSDATE is faster as it doesn't need time zone conversion
-- Use SYSDATE for logging and timestamps unless time zone matters
-- Use CURRENT_DATE when user's time zone is important

-- Clean up
DROP TABLE event_log;

Output / Afterresult

After

COUNT	AVG_SALARY	MIN_SALARY	MAX_SALARY	SUM_SALARY
4	59,250	50,000	70,000	237,000

Aggregate over all 4 employees — AVG=(50+62+55+70)/4=59,250

Key Takeaway

SYSDATE returns the current date and time of the database server, while CURRENT_DATE returns the current date and time in the session time zone.

Q141. Write a query to get the length of a string in a column in Oracle.

Answer

Use the LENGTH function to get the number of characters in a string, or LENGTHB for byte length.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

CodeOracle SQL

SQL

-- Basic LENGTH function
SELECT first_name,
       LENGTH(first_name) as name_length
FROM employees;

-- LENGTH vs LENGTHB (character vs byte length)
SELECT first_name,
       LENGTH(first_name) as char_length,
       LENGTHB(first_name) as byte_length
FROM employees;

-- Find employees with names longer than 8 characters
SELECT first_name, last_name,
       LENGTH(first_name) as first_name_length,
       LENGTH(last_name) as last_name_length
FROM employees
WHERE LENGTH(first_name) > 8;

-- Combined string length
SELECT first_name,
       last_name,
       LENGTH(first_name || ' ' || last_name) as full_name_length
FROM employees;

-- Handle NULL values
SELECT first_name,
       middle_name,
       LENGTH(first_name) as first_length,
       LENGTH(middle_name) as middle_length,
       NVL(LENGTH(middle_name), 0) as middle_length_safe
FROM employees;

-- Statistical analysis of string lengths
SELECT MIN(LENGTH(first_name)) as min_length,
       MAX(LENGTH(first_name)) as max_length,
       AVG(LENGTH(first_name)) as avg_length,
       STDDEV(LENGTH(first_name)) as std_dev
FROM employees;

-- Group by string length
SELECT LENGTH(first_name) as name_length,
       COUNT(*) as count_of_names
FROM employees
GROUP BY LENGTH(first_name)
ORDER BY name_length;

-- Find longest and shortest names
SELECT first_name,
       LENGTH(first_name) as name_length
FROM employees
WHERE LENGTH(first_name) = (SELECT MAX(LENGTH(first_name)) FROM employees)
   OR LENGTH(first_name) = (SELECT MIN(LENGTH(first_name)) FROM employees);

-- Length-based filtering and sorting
SELECT first_name,
       last_name,
       LENGTH(first_name) as first_length,
       LENGTH(last_name) as last_length
FROM employees
WHERE LENGTH(first_name) BETWEEN 4 AND 8
ORDER BY LENGTH(first_name) DESC, first_name;

-- Character vs byte length for multibyte characters
-- (Important for Unicode/UTF-8 data)
SELECT 'Hello' as text,
       LENGTH('Hello') as char_length,
       LENGTHB('Hello') as byte_length
FROM dual
UNION ALL
SELECT 'Héllo' as text,
       LENGTH('Héllo') as char_length,
       LENGTHB('Héllo') as byte_length
FROM dual;

-- Practical example: Data validation
SELECT email,
       LENGTH(email) as email_length,
       CASE 
           WHEN LENGTH(email) > 100 THEN 'Too Long'
           WHEN LENGTH(email) < 5 THEN 'Too Short'
           WHEN email NOT LIKE '%@%' THEN 'Invalid Format'
           ELSE 'Valid'
       END as validation_status
FROM employees;

-- Find records with empty or very short values
SELECT employee_id,
       first_name,
       LENGTH(first_name) as length
FROM employees
WHERE LENGTH(first_name) <= 2 OR first_name IS NULL;

-- Length comparison across columns
SELECT employee_id,
       first_name,
       last_name,
       LENGTH(first_name) as first_len,
       LENGTH(last_name) as last_len,
       CASE 
           WHEN LENGTH(first_name) > LENGTH(last_name) THEN 'First name longer'
           WHEN LENGTH(first_name) < LENGTH(last_name) THEN 'Last name longer'
           ELSE 'Same length'
       END as comparison
FROM employees;

-- Substring and length operations
SELECT first_name,
       SUBSTR(first_name, 1, 3) as first_three,
       LENGTH(SUBSTR(first_name, 1, 3)) as substr_length,
       LENGTH(TRIM(first_name)) as trimmed_length
FROM employees;

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

Data state before query executes.

Output / Afterresult

After

EMP_ID	EMP_NAME	SOURCE
101	Alice Johnson	CURRENT
102	Bob Smith	CURRENT
101	Alice Johnson	ARCHIVE
102	Bob Smith	ARCHIVE

4 rows — UNION ALL keeps duplicates, faster than UNION

Key Takeaway

Use the LENGTH function to get the number of characters in a string, or LENGTHB for byte length.

Q142. What is the purpose of the LISTAGG function in Oracle?

Answer

LISTAGG is an analytical function that concatenates values from multiple rows into a single string, with optional ordering and separator characters.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

CodeOracle SQL

SQL

-- Basic LISTAGG usage
SELECT department_id,
       LISTAGG(first_name, ', ') WITHIN GROUP (ORDER BY first_name) as employee_names
FROM employees
GROUP BY department_id;

-- LISTAGG with custom separator
SELECT department_id,
       LISTAGG(first_name, ' | ') WITHIN GROUP (ORDER BY hire_date) as employees_by_hire_date
FROM employees
GROUP BY department_id;

-- LISTAGG with complex expressions
SELECT department_id,
       LISTAGG(first_name || ' (' || job_id || ')', '; ') 
       WITHIN GROUP (ORDER BY salary DESC) as employee_details
FROM employees
GROUP BY department_id;

-- LISTAGG as window function (Oracle 12c+)
SELECT employee_id,
       first_name,
       department_id,
       LISTAGG(first_name, ', ') WITHIN GROUP (ORDER BY first_name) 
       OVER (PARTITION BY department_id) as dept_colleagues
FROM employees;

-- Handle potential overflow with ON OVERFLOW TRUNCATE (Oracle 12c+)
SELECT department_id,
       LISTAGG(first_name, ', ' ON OVERFLOW TRUNCATE '...' WITH COUNT) 
       WITHIN GROUP (ORDER BY first_name) as employee_names
FROM employees
GROUP BY department_id;

-- LISTAGG with DISTINCT (Oracle 19c+)
SELECT department_id,
       LISTAGG(DISTINCT job_id, ', ') WITHIN GROUP (ORDER BY job_id) as unique_jobs
FROM employees
GROUP BY department_id;

-- Practical example: Create employee directory
SELECT d.department_name,
       COUNT(e.employee_id) as employee_count,
       LISTAGG(e.first_name || ' ' || e.last_name, CHR(10)) 
       WITHIN GROUP (ORDER BY e.last_name) as employee_list
FROM departments d
LEFT JOIN employees e ON d.department_id = e.department_id
GROUP BY d.department_name
ORDER BY d.department_name;

-- LISTAGG with conditional logic
SELECT department_id,
       LISTAGG(
           CASE 
               WHEN salary > 50000 THEN first_name || ' (High)'
               ELSE first_name || ' (Standard)'
           END, 
           ', '
       ) WITHIN GROUP (ORDER BY salary DESC) as employee_salary_categories
FROM employees
GROUP BY department_id;

-- Alternative to LISTAGG for older Oracle versions
SELECT department_id,
       SUBSTR(
           SYS_CONNECT_BY_PATH(first_name, ','), 2
       ) as employee_names
FROM (
    SELECT department_id, first_name,
           ROW_NUMBER() OVER (PARTITION BY department_id ORDER BY first_name) as rn,
           COUNT(*) OVER (PARTITION BY department_id) as cnt
    FROM employees
)
WHERE rn = cnt
START WITH rn = 1
CONNECT BY PRIOR rn = rn - 1 AND PRIOR department_id = department_id;

-- Using XMLAGG as alternative
SELECT department_id,
       RTRIM(
           XMLAGG(
               XMLELEMENT("x", first_name || ', ')
           ).getClobVal(), 
           ', '
       ) as employee_names
FROM employees
GROUP BY department_id;

-- LISTAGG with hierarchical data
SELECT LEVEL,
       LPAD(' ', (LEVEL-1)*2) || first_name as hierarchy_display,
       SYS_CONNECT_BY_PATH(first_name, ' -> ') as management_chain
FROM employees
START WITH manager_id IS NULL
CONNECT BY PRIOR employee_id = manager_id;

-- LISTAGG for creating dynamic SQL
SELECT 'SELECT ' || 
       LISTAGG(column_name, ', ') WITHIN GROUP (ORDER BY column_id) ||
       ' FROM ' || table_name as dynamic_select
FROM user_tab_columns
WHERE table_name = 'EMPLOYEES'
GROUP BY table_name;

-- Performance considerations and limitations
-- LISTAGG has a 4000 character limit in older versions
-- Use CLOB version for larger strings (Oracle 12c+)
SELECT department_id,
       LISTAGG(first_name || ' ' || last_name, ', ') 
       WITHIN GROUP (ORDER BY first_name) as employee_names_clob
FROM employees
GROUP BY department_id;

-- Clean up demonstration tables
-- (No tables created in this example)

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

Data state before query executes.

Output / Afterresult

After

EMPLOYEE	MANAGER
Bob Smith	Alice Johnson
Carol White	Alice Johnson
Dave Brown	Bob Smith

3 rows — self join pairs each employee with their manager's name

Key Takeaway

LISTAGG is an analytical function that concatenates values from multiple rows into a single string, with optional ordering and separator characters.

Q143. How do you convert the data type of a column in an Oracle table?

Answer

Use ALTER TABLE MODIFY statement to change the data type of a column, with considerations for data compatibility and potential data loss.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

CodeOracle SQL

SQL

-- Basic syntax for modifying column data type
ALTER TABLE employees MODIFY (salary NUMBER(10,2));

-- Create test table for demonstration
CREATE TABLE data_type_demo (
    id NUMBER,
    text_col VARCHAR2(50),
    number_col NUMBER,
    date_col DATE
);

INSERT INTO data_type_demo VALUES (1, '12345', 100, SYSDATE);
INSERT INTO data_type_demo VALUES (2, 'Hello', 200, SYSDATE-1);

-- Example 1: Increase VARCHAR2 size
ALTER TABLE data_type_demo MODIFY (text_col VARCHAR2(100));

-- Example 2: Change NUMBER precision
ALTER TABLE data_type_demo MODIFY (number_col NUMBER(10,2));

-- Example 3: Converting compatible types
-- VARCHAR2 to NUMBER (data must be numeric)
UPDATE data_type_demo SET text_col = '999' WHERE id = 2;
ALTER TABLE data_type_demo MODIFY (text_col NUMBER);

-- Reverting back for next example
ALTER TABLE data_type_demo MODIFY (text_col VARCHAR2(100));
UPDATE data_type_demo SET text_col = 'Hello' WHERE id = 2;

-- Example 4: Converting with data transformation
-- Step 1: Add new column with desired type
ALTER TABLE data_type_demo ADD (new_text_col CLOB);

-- Step 2: Copy and transform data
UPDATE data_type_demo SET new_text_col = text_col;

-- Step 3: Drop old column
ALTER TABLE data_type_demo DROP COLUMN text_col;

-- Step 4: Rename new column
ALTER TABLE data_type_demo RENAME COLUMN new_text_col TO text_col;

-- Example 5: Safe conversion process for complex changes
-- Create backup table
CREATE TABLE data_type_demo_backup AS SELECT * FROM data_type_demo;

-- Complex conversion: VARCHAR2 to DATE
ALTER TABLE data_type_demo ADD (temp_date_col DATE);

-- Convert data with proper handling
UPDATE data_type_demo 
SET temp_date_col = CASE 
    WHEN REGEXP_LIKE(text_col, '^\d{4}-\d{2}-\d{2}) 
    THEN TO_DATE(text_col, 'YYYY-MM-DD')
    ELSE NULL 
END;

-- Check conversion results
SELECT text_col, temp_date_col FROM data_type_demo;

-- Complete the conversion
ALTER TABLE data_type_demo DROP COLUMN text_col;
ALTER TABLE data_type_demo RENAME COLUMN temp_date_col TO text_col;

-- Example 6: Multiple column modifications
ALTER TABLE data_type_demo MODIFY (
    id NUMBER(10),
    number_col NUMBER(15,3),
    date_col TIMESTAMP
);

-- Example 7: Adding constraints during modification
ALTER TABLE data_type_demo MODIFY (
    id NUMBER(10) NOT NULL,
    number_col NUMBER(15,3) DEFAULT 0
);

-- Example 8: Handling constraints during type changes
-- View existing constraints
SELECT constraint_name, constraint_type, column_name
FROM user_cons_columns
WHERE table_name = 'DATA_TYPE_DEMO';

-- Disable constraints if needed
-- ALTER TABLE data_type_demo DISABLE CONSTRAINT constraint_name;

-- Modify column
-- ALTER TABLE data_type_demo MODIFY (column_name NEW_TYPE);

-- Re-enable constraints
-- ALTER TABLE data_type_demo ENABLE CONSTRAINT constraint_name;

-- Example 9: Check data compatibility before conversion
-- Check if all values in number_col can be converted to INTEGER
SELECT COUNT(*) as total_rows,
       COUNT(CASE WHEN MOD(number_col, 1) = 0 THEN 1 END) as integer_compatible
FROM data_type_demo;

-- Example 10: Using CAST for validation
SELECT text_col,
       CASE 
           WHEN text_col IS NULL THEN 'NULL'
           WHEN REGEXP_LIKE(text_col, '^\d+) THEN 'NUMERIC'
           WHEN REGEXP_LIKE(text_col, '^\d{4}-\d{2}-\d{2}) THEN 'DATE_FORMAT'
           ELSE 'TEXT'
       END as data_type_category
FROM data_type_demo;

-- View current column definitions
SELECT column_name, data_type, data_length, data_precision, data_scale, nullable
FROM user_tab_columns
WHERE table_name = 'DATA_TYPE_DEMO'
ORDER BY column_id;

-- Common conversion scenarios and best practices:

-- 1. VARCHAR2 size increase: Always safe
-- ALTER TABLE table_name MODIFY (column_name VARCHAR2(new_larger_size));

-- 2. VARCHAR2 size decrease: Check data first
-- SELECT MAX(LENGTH(column_name)) FROM table_name;

-- 3. NUMBER precision increase: Usually safe
-- ALTER TABLE table_name MODIFY (column_name NUMBER(larger_precision, scale));

-- 4. NUMBER to VARCHAR2: Usually safe
-- ALTER TABLE table_name MODIFY (column_name VARCHAR2(sufficient_size));

-- 5. VARCHAR2 to NUMBER: Validate data first
-- SELECT column_name FROM table_name WHERE NOT REGEXP_LIKE(column_name, '^-?\d+(\.\d+)?);

-- 6. DATE to VARCHAR2: Safe with proper formatting
-- ALTER TABLE table_name ADD (temp_col VARCHAR2(19));
-- UPDATE table_name SET temp_col = TO_CHAR(date_col, 'YYYY-MM-DD HH24:MI:SS');

-- 7. VARCHAR2 to DATE: Validate format first
-- SELECT COUNT(*) FROM table_name WHERE column_name IS NOT NULL 
-- AND NOT REGEXP_LIKE(column_name, '^\d{4}-\d{2}-\d{2}');

-- Clean up
DROP TABLE data_type_demo;
DROP TABLE data_type_demo_backup;

-- Important notes:
-- 1. Always backup data before major conversions
-- 2. Test conversions on development environment first
-- 3. Check for data loss during precision reduction
-- 4. Consider impact on indexes and constraints
-- 5. Update application code if data types change
-- 6. Some conversions may require table recreation

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

Data state before query executes.

Output / Afterresult

After

COUNT	AVG_SALARY	MIN_SALARY	MAX_SALARY	SUM_SALARY
4	59,250	50,000	70,000	237,000

Aggregate over all 4 employees — AVG=(50+62+55+70)/4=59,250

Key Takeaway

Use ALTER TABLE MODIFY statement to change the data type of a column, with considerations for data compatibility and potential data loss.

Q144. How would you retrieve the total count of rows, but only count each distinct value in a column once in Oracle?

Answer

Use COUNT(DISTINCT column_name) to count unique values, or combine COUNT with DISTINCT in subqueries for more complex scenarios.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

CodeOracle SQL

SQL

-- Basic COUNT DISTINCT
SELECT COUNT(DISTINCT department_id) as unique_departments
FROM employees;

-- Count distinct values across multiple columns
SELECT COUNT(DISTINCT department_id) as unique_departments,
       COUNT(DISTINCT job_id) as unique_jobs,
       COUNT(DISTINCT manager_id) as unique_managers
FROM employees;

-- Count distinct combinations of multiple columns
SELECT COUNT(DISTINCT department_id || '_' || job_id) as unique_dept_job_combinations
FROM employees;

-- Alternative using GROUP BY for distinct count
SELECT COUNT(*) as unique_departments
FROM (
    SELECT DISTINCT department_id
    FROM employees
);

-- Count distinct with conditions
SELECT COUNT(DISTINCT CASE WHEN salary > 50000 THEN department_id END) as high_salary_departments
FROM employees;

-- Compare total vs distinct counts
SELECT COUNT(*) as total_employees,
       COUNT(DISTINCT department_id) as unique_departments,
       COUNT(DISTINCT job_id) as unique_jobs,
       COUNT(*) / COUNT(DISTINCT department_id) as avg_employees_per_dept
FROM employees;

-- Distinct count by group
SELECT department_id,
       COUNT(*) as total_employees,
       COUNT(DISTINCT job_id) as unique_jobs_in_dept
FROM employees
GROUP BY department_id;

-- Multiple distinct counts with filtering
SELECT COUNT(DISTINCT employee_id) as unique_employees,
       COUNT(DISTINCT CASE WHEN salary > 50000 THEN employee_id END) as high_earners,
       COUNT(DISTINCT CASE WHEN hire_date >= ADD_MONTHS(SYSDATE, -12) THEN employee_id END) as recent_hires
FROM employees;

-- Using analytical functions for distinct counts
SELECT department_id,
       employee_id,
       COUNT(DISTINCT department_id) OVER () as total_unique_departments,
       COUNT(DISTINCT job_id) OVER (PARTITION BY department_id) as unique_jobs_in_current_dept
FROM employees;

-- Distinct count with NULL handling
SELECT COUNT(DISTINCT NVL(commission_pct, -1)) as unique_commission_rates,
       COUNT(DISTINCT commission_pct) as unique_non_null_commission_rates
FROM employees;

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

Data state before query executes.

Output / Afterresult

After

EMPLOYEE	MANAGER
Bob Smith	Alice Johnson
Carol White	Alice Johnson
Dave Brown	Bob Smith

3 rows — self join pairs each employee with their manager's name

Key Takeaway

Use COUNT(DISTINCT column_name) to count unique values, or combine COUNT with DISTINCT in subqueries for more complex scenarios.

Q145. How would you find the three most frequent values in a column along with their counts in Oracle?

Answer

Use GROUP BY with COUNT and ORDER BY, combined with ROWNUM or analytical functions to limit results to top 3.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

CodeOracle SQL

SQL

-- Method 1: Using ROWNUM (Oracle 11g and earlier)
SELECT * FROM (
    SELECT job_id, COUNT(*) as frequency
    FROM employees
    GROUP BY job_id
    ORDER BY COUNT(*) DESC
)
WHERE ROWNUM <= 3;

-- Method 2: Using ROW_NUMBER() analytical function
SELECT job_id, frequency FROM (
    SELECT job_id, 
           COUNT(*) as frequency,
           ROW_NUMBER() OVER (ORDER BY COUNT(*) DESC) as rn
    FROM employees
    GROUP BY job_id
)
WHERE rn <= 3;

-- Method 3: Using RANK() to handle ties
SELECT job_id, frequency FROM (
    SELECT job_id, 
           COUNT(*) as frequency,
           RANK() OVER (ORDER BY COUNT(*) DESC) as rank_num
    FROM employees
    GROUP BY job_id
)
WHERE rank_num <= 3;

-- Method 4: Using FETCH FIRST (Oracle 12c+)
SELECT job_id, COUNT(*) as frequency
FROM employees
GROUP BY job_id
ORDER BY COUNT(*) DESC
FETCH FIRST 3 ROWS ONLY;

-- Method 5: Including percentage and cumulative statistics
SELECT job_id, 
       frequency,
       ROUND(frequency * 100.0 / total_count, 2) as percentage,
       ROUND(SUM(frequency) OVER (ORDER BY frequency DESC) * 100.0 / total_count, 2) as cumulative_percentage
FROM (
    SELECT job_id, 
           COUNT(*) as frequency,
           SUM(COUNT(*)) OVER () as total_count,
           ROW_NUMBER() OVER (ORDER BY COUNT(*) DESC) as rn
    FROM employees
    GROUP BY job_id
)
WHERE rn <= 3;

-- Method 6: Most frequent values by department
SELECT department_id, job_id, frequency FROM (
    SELECT department_id,
           job_id, 
           COUNT(*) as frequency,
           ROW_NUMBER() OVER (PARTITION BY department_id ORDER BY COUNT(*) DESC) as rn
    FROM employees
    GROUP BY department_id, job_id
)
WHERE rn <= 3
ORDER BY department_id, frequency DESC;

-- Method 7: Using DENSE_RANK for better tie handling
SELECT job_id, frequency, rank_position FROM (
    SELECT job_id, 
           COUNT(*) as frequency,
           DENSE_RANK() OVER (ORDER BY COUNT(*) DESC) as rank_position
    FROM employees
    GROUP BY job_id
)
WHERE rank_position <= 3
ORDER BY rank_position, job_id;

-- Method 8: Most frequent salary ranges
SELECT salary_range, frequency FROM (
    SELECT CASE 
               WHEN salary < 30000 THEN 'Low (< 30K)'
               WHEN salary < 60000 THEN 'Medium (30K-60K)'
               WHEN salary < 100000 THEN 'High (60K-100K)'
               ELSE 'Very High (> 100K)'
           END as salary_range,
           COUNT(*) as frequency,
           ROW_NUMBER() OVER (ORDER BY COUNT(*) DESC) as rn
    FROM employees
    GROUP BY CASE 
                 WHEN salary < 30000 THEN 'Low (< 30K)'
                 WHEN salary < 60000 THEN 'Medium (30K-60K)'
                 WHEN salary < 100000 THEN 'High (60K-100K)'
                 ELSE 'Very High (> 100K)'
             END
)
WHERE rn <= 3;

-- Method 9: Including additional statistics
SELECT job_id,
       frequency,
       ROUND(frequency * 100.0 / SUM(frequency) OVER (), 2) as percentage,
       ROUND(AVG(frequency) OVER (), 2) as avg_frequency,
       frequency - ROUND(AVG(frequency) OVER (), 2) as deviation_from_avg
FROM (
    SELECT job_id, COUNT(*) as frequency
    FROM employees
    GROUP BY job_id
    ORDER BY COUNT(*) DESC
)
WHERE ROWNUM <= 3;

-- Method 10: Dynamic top N using PL/SQL
DECLARE
    v_top_n NUMBER := 3;
    v_sql VARCHAR2(1000);
    TYPE freq_record IS RECORD (
        job_id VARCHAR2(10),
        frequency NUMBER
    );
    TYPE freq_table IS TABLE OF freq_record;
    v_results freq_table;
BEGIN
    v_sql := 'SELECT job_id, frequency FROM (
                SELECT job_id, COUNT(*) as frequency,
                       ROW_NUMBER() OVER (ORDER BY COUNT(*) DESC) as rn
                FROM employees
                GROUP BY job_id
              ) WHERE rn <= :1';
    
    EXECUTE IMMEDIATE v_sql BULK COLLECT INTO v_results USING v_top_n;
    
    DBMS_OUTPUT.PUT_LINE('Top ' || v_top_n || ' most frequent job IDs:');
    FOR i IN 1..v_results.COUNT LOOP
        DBMS_OUTPUT.PUT_LINE(v_results(i).job_id || ': ' || v_results(i).frequency);
    END LOOP;
END;
/

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

Data state before query executes.

Output / Afterresult

After

OPERATION	ROWS	CONTEXT_SWITCHES
Row-by-row	4	4 (slow)
BULK COLLECT	4	1 (fast)

BULK COLLECT fetches all rows at once — reduces SQL↔PL/SQL round trips

Key Takeaway

Use GROUP BY with COUNT and ORDER BY, combined with ROWNUM or analytical functions to limit results to top 3.

Q146. Write a query to get the monthly sales amount for the last 12 months in Oracle.

Q147. Write a query to find the median salary for each department without using built-in median functions.

Answer

Calculate median using window functions and percentile logic by finding the middle value(s) in ordered datasets.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

CodeOracle SQL

SQL

-- Method 1: Using PERCENTILE_CONT (if allowed)
SELECT department_id,
      PERCENTILE_CONT(0.5) WITHIN GROUP (ORDER BY salary) as median_salary
FROM employees
GROUP BY department_id;

-- Method 2: Manual median calculation using ROW_NUMBER
WITH ranked_salaries AS (
   SELECT department_id, salary,
          ROW_NUMBER() OVER (PARTITION BY department_id ORDER BY salary) as rn,
          COUNT(*) OVER (PARTITION BY department_id) as total_count
   FROM employees
),
median_positions AS (
   SELECT department_id, salary, rn, total_count,
          CASE 
              WHEN MOD(total_count, 2) = 1 THEN (total_count + 1) / 2
              ELSE total_count / 2
          END as median_pos1,

          CASE 
              WHEN MOD(total_count, 2) = 1 THEN (total_count + 1) / 2
              ELSE (total_count / 2) + 1
          END as median_pos2
   FROM ranked_salaries
)
SELECT department_id,
      CASE 
          WHEN total_count = 1 THEN 
              MAX(CASE WHEN rn = median_pos1 THEN salary END)
          WHEN MOD(total_count, 2) = 1 THEN 
              MAX(CASE WHEN rn = median_pos1 THEN salary END)
          ELSE 
              (MAX(CASE WHEN rn = median_pos1 THEN salary END) + 
               MAX(CASE WHEN rn = median_pos2 THEN salary END)) / 2
      END as median_salary
FROM median_positions
GROUP BY department_id, total_count;

-- Method 3: Using NTILE for approximate median
WITH salary_quartiles AS (
   SELECT department_id, salary,
          NTILE(2) OVER (PARTITION BY department_id ORDER BY salary) as half
   FROM employees
)
SELECT department_id,
      AVG(salary) as approximate_median
FROM salary_quartiles
WHERE half = 1
GROUP BY department_id;

-- Method 4: Using analytical functions with proper median logic
WITH ordered_salaries AS (
   SELECT department_id, salary,
          ROW_NUMBER() OVER (PARTITION BY department_id ORDER BY salary ASC) as rn_asc,
          ROW_NUMBER() OVER (PARTITION BY department_id ORDER BY salary DESC) as rn_desc,
          COUNT(*) OVER (PARTITION BY department_id) as cnt
   FROM employees
)
SELECT department_id,
      AVG(salary) as median_salary
FROM ordered_salaries
WHERE rn_asc IN (FLOOR((cnt + 1) / 2), CEIL((cnt + 1) / 2))
GROUP BY department_id;

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

Data state before query executes.

Output / Afterresult

After

EMP_NAME	SALARY	ROW_NUM	RANK	DENSE_RANK
Dave Brown	70,000	1	1	1
Bob Smith	62,000	2	2	2
Carol White	55,000	3	3	3
Alice Johnson	50,000	4	4	4

All 4 employees ranked by salary DESC — no ties so all methods agree

Key Takeaway

Calculate median using window functions and percentile logic by finding the middle value(s) in ordered datasets.

Q148. Write a query to detect and remove duplicate records while preserving the most recent version.

Answer

Use window functions to identify duplicates and remove older versions while keeping the most recent record based on timestamp or other criteria.

CodeOracle SQL

SQL

-- Sample table with duplicates
CREATE TABLE customer_records (
   record_id NUMBER,
   customer_id NUMBER,
   customer_name VARCHAR2(100),
   email VARCHAR2(100),
   created_date DATE,
   last_updated DATE
);

-- Method 1: Identify duplicates using ROW_NUMBER()
WITH duplicate_analysis AS (
   SELECT record_id, customer_id, customer_name, email, 
          created_date, last_updated,
          ROW_NUMBER() OVER (
              PARTITION BY customer_id 
              ORDER BY last_updated DESC, record_id DESC
          ) as rn,
          COUNT(*) OVER (PARTITION BY customer_id) as duplicate_count
   FROM customer_records
)
SELECT record_id, customer_id, customer_name, 
      CASE WHEN rn = 1 THEN 'KEEP' ELSE 'DELETE' END as action,
      duplicate_count
FROM duplicate_analysis
WHERE duplicate_count > 1
ORDER BY customer_id, rn;

-- Method 2: Delete duplicates keeping most recent
DELETE FROM customer_records
WHERE record_id IN (
   SELECT record_id
   FROM (
       SELECT record_id,
              ROW_NUMBER() OVER (
                  PARTITION BY customer_id 
                  ORDER BY last_updated DESC, record_id DESC
              ) as rn
       FROM customer_records
   )
   WHERE rn > 1
);

-- Method 3: Using MERGE to handle duplicates during insert
MERGE INTO customer_master cm
USING (
   SELECT customer_id, customer_name, email, last_updated,
          ROW_NUMBER() OVER (
              PARTITION BY customer_id 
              ORDER BY last_updated DESC
          ) as rn
   FROM customer_staging
) cs ON (cm.customer_id = cs.customer_id AND cs.rn = 1)
WHEN MATCHED THEN
   UPDATE SET 
       customer_name = cs.customer_name,
       email = cs.email,
       last_updated = cs.last_updated
   WHERE cs.last_updated > cm.last_updated
WHEN NOT MATCHED THEN
   INSERT (customer_id, customer_name, email, last_updated)
   VALUES (cs.customer_id, cs.customer_name, cs.email, cs.last_updated);

-- Method 4: Create deduplicated table
CREATE TABLE customer_clean AS
SELECT record_id, customer_id, customer_name, email, 
      created_date, last_updated
FROM (
   SELECT record_id, customer_id, customer_name, email, 
          created_date, last_updated,
          ROW_NUMBER() OVER (
              PARTITION BY customer_id 
              ORDER BY last_updated DESC, record_id DESC
          ) as rn
   FROM customer_records
)
WHERE rn = 1;

-- Method 5: Complex deduplication with business rules
WITH advanced_dedup AS (
   SELECT record_id, customer_id, customer_name, email, 
          created_date, last_updated,
          -- Priority: most recent update, then most complete record, then highest ID
          ROW_NUMBER() OVER (
              PARTITION BY customer_id 
              ORDER BY 
                  last_updated DESC,
                  CASE WHEN email IS NOT NULL THEN 1 ELSE 2 END,
                  LENGTH(customer_name) DESC,
                  record_id DESC
          ) as priority_rank,
          COUNT(*) OVER (PARTITION BY customer_id) as total_duplicates
   FROM customer_records
)
SELECT customer_id, customer_name, email, last_updated,
      'KEEP - Priority: ' || priority_rank as status
FROM advanced_dedup
WHERE priority_rank = 1;

-- Method 6: Duplicate detection with EXCEPTION handling
CREATE OR REPLACE PROCEDURE clean_customer_duplicates IS
   v_deleted_count NUMBER := 0;
BEGIN
   -- Create backup first
   EXECUTE IMMEDIATE 'CREATE TABLE customer_backup_' || 
                    TO_CHAR(SYSDATE, 'YYYYMMDDHH24MISS') || 
                    ' AS SELECT * FROM customer_records';
   
   -- Delete duplicates
   DELETE FROM customer_records
   WHERE record_id IN (
       SELECT record_id
       FROM (
           SELECT record_id,
                  ROW_NUMBER() OVER (
                      PARTITION BY customer_id 
                      ORDER BY last_updated DESC, record_id DESC
                  ) as rn
           FROM customer_records
       )
       WHERE rn > 1
   );
   
   v_deleted_count := SQL%ROWCOUNT;
   
   DBMS_OUTPUT.PUT_LINE('Deleted ' || v_deleted_count || ' duplicate records');
   
   COMMIT;
EXCEPTION
   WHEN OTHERS THEN
       ROLLBACK;
       DBMS_OUTPUT.PUT_LINE('Error during deduplication: ' || SQLERRM);
       RAISE;
END;
/

-- Method 7: Report duplicate statistics before cleanup
SELECT customer_id,
      COUNT(*) as total_records,
      COUNT(*) - 1 as duplicates_to_remove,
      MIN(created_date) as first_created,
      MAX(last_updated) as most_recent_update,
      LISTAGG(record_id, ',') WITHIN GROUP (ORDER BY last_updated DESC) as all_record_ids
FROM customer_records
GROUP BY customer_id
HAVING COUNT(*) > 1
ORDER BY COUNT(*) DESC;

Output / Afterresult

After

EMP_ID	EMP_NAME	ACTION	RESULT
101	Alice Johnson	UPDATED (matched)	Salary → 55,000
102	Bob Smith	UPDATED (matched)	Salary → 68,200
999	New Hire	INSERTED (not matched)	New row added

MERGE: 2 rows updated + 1 inserted in one atomic statement

Key Takeaway

Use window functions to identify duplicates and remove older versions while keeping the most recent record based on timestamp or other criteria.

Q149. Write a query to calculate year-over-year growth percentage for sales data.

Answer

Use LAG window function or self-joins to compare current period values with previous period values and calculate growth percentages.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

CodeOracle SQL

SQL

-- Sample sales data setup
CREATE TABLE monthly_sales (
   year_month DATE,
   sales_amount NUMBER,
   region VARCHAR2(50)
);

-- Method 1: Using LAG window function
SELECT year_month,
      sales_amount,
      LAG(sales_amount, 12) OVER (ORDER BY year_month) as previous_year_sales,
      sales_amount - LAG(sales_amount, 12) OVER (ORDER BY year_month) as absolute_growth,
      ROUND(
          ((sales_amount - LAG(sales_amount, 12) OVER (ORDER BY year_month)) / 
           LAG(sales_amount, 12) OVER (ORDER BY year_month)) * 100, 2
      ) as yoy_growth_percentage
FROM monthly_sales
ORDER BY year_month;

-- Method 2: YoY growth by region
SELECT region, year_month, sales_amount,
      LAG(sales_amount, 12) OVER (PARTITION BY region ORDER BY year_month) as prev_year_sales,
      CASE 
          WHEN LAG(sales_amount, 12) OVER (PARTITION BY region ORDER BY year_month) IS NOT NULL
          THEN ROUND(
              ((sales_amount - LAG(sales_amount, 12) OVER (PARTITION BY region ORDER BY year_month)) / 
               LAG(sales_amount, 12) OVER (PARTITION BY region ORDER BY year_month)) * 100, 2
          )
          ELSE NULL
      END as yoy_growth_pct
FROM monthly_sales
ORDER BY region, year_month;

-- Method 3: Annual aggregated YoY growth
WITH annual_sales AS (
   SELECT EXTRACT(YEAR FROM year_month) as sales_year,
          SUM(sales_amount) as total_sales
   FROM monthly_sales
   GROUP BY EXTRACT(YEAR FROM year_month)
)
SELECT sales_year,
      total_sales,
      LAG(total_sales) OVER (ORDER BY sales_year) as previous_year_total,
      total_sales - LAG(total_sales) OVER (ORDER BY sales_year) as absolute_growth,
      ROUND(
          ((total_sales - LAG(total_sales) OVER (ORDER BY sales_year)) / 
           LAG(total_sales) OVER (ORDER BY sales_year)) * 100, 2
      ) as yoy_growth_percentage
FROM annual_sales
ORDER BY sales_year;

-- Method 4: Quarterly YoY comparison
WITH quarterly_sales AS (
   SELECT EXTRACT(YEAR FROM year_month) as sales_year,
          CEIL(EXTRACT(MONTH FROM year_month) / 3) as quarter,
          SUM(sales_amount) as quarterly_sales
   FROM monthly_sales
   GROUP BY EXTRACT(YEAR FROM year_month), CEIL(EXTRACT(MONTH FROM year_month) / 3)
)
SELECT sales_year, quarter, quarterly_sales,
      LAG(quarterly_sales, 4) OVER (PARTITION BY quarter ORDER BY sales_year) as prev_year_quarter,
      ROUND(
          CASE 
              WHEN LAG(quarterly_sales, 4) OVER (PARTITION BY quarter ORDER BY sales_year) > 0
              THEN ((quarterly_sales - LAG(quarterly_sales, 4) OVER (PARTITION BY quarter ORDER BY sales_year)) / 
                    LAG(quarterly_sales, 4) OVER (PARTITION BY quarter ORDER BY sales_year)) * 100
              ELSE NULL
          END, 2
      ) as yoy_quarter_growth_pct
FROM quarterly_sales
ORDER BY quarter, sales_year;

-- Method 5: Self-join approach (alternative to LAG)
SELECT this_year.year_month,
      this_year.sales_amount as current_sales,
      last_year.sales_amount as previous_year_sales,
      ROUND(
          ((this_year.sales_amount - last_year.sales_amount) / last_year.sales_amount) * 100, 2
      ) as yoy_growth_percentage
FROM monthly_sales this_year
LEFT JOIN monthly_sales last_year 
   ON ADD_MONTHS(last_year.year_month, 12) = this_year.year_month
   AND this_year.region = last_year.region
ORDER BY this_year.year_month;

-- Method 6: Multi-period comparison (YoY, QoQ, MoM)
SELECT year_month, sales_amount,
      -- Year over Year
      LAG(sales_amount, 12) OVER (ORDER BY year_month) as yoy_prev,
      ROUND(((sales_amount - LAG(sales_amount, 12) OVER (ORDER BY year_month)) / 
             NULLIF(LAG(sales_amount, 12) OVER (ORDER BY year_month), 0)) * 100, 2) as yoy_growth,
      
      -- Quarter over Quarter (3 months)
      LAG(sales_amount, 3) OVER (ORDER BY year_month) as qoq_prev,
      ROUND(((sales_amount - LAG(sales_amount, 3) OVER (ORDER BY year_month)) / 
             NULLIF(LAG(sales_amount, 3) OVER (ORDER BY year_month), 0)) * 100, 2) as qoq_growth,
      
      -- Month over Month
      LAG(sales_amount, 1) OVER (ORDER BY year_month) as mom_prev,
      ROUND(((sales_amount - LAG(sales_amount, 1) OVER (ORDER BY year_month)) / 
             NULLIF(LAG(sales_amount, 1) OVER (ORDER BY year_month), 0)) * 100, 2) as mom_growth
FROM monthly_sales
ORDER BY year_month;

-- Method 7: Growth categorization
WITH growth_analysis AS (
   SELECT year_month, sales_amount, region,
          LAG(sales_amount, 12) OVER (PARTITION BY region ORDER BY year_month) as prev_year_sales,
          ROUND(
              ((sales_amount - LAG(sales_amount, 12) OVER (PARTITION BY region ORDER BY year_month)) / 
               NULLIF(LAG(sales_amount, 12) OVER (PARTITION BY region ORDER BY year_month), 0)) * 100, 2
          ) as yoy_growth_pct
   FROM monthly_sales
)
SELECT year_month, region, sales_amount, prev_year_sales, yoy_growth_pct,
      CASE 
          WHEN yoy_growth_pct IS NULL THEN 'No Prior Data'
          WHEN yoy_growth_pct >= 20 THEN 'High Growth'
          WHEN yoy_growth_pct >= 10 THEN 'Moderate Growth'
          WHEN yoy_growth_pct >= 0 THEN 'Low Growth'
          WHEN yoy_growth_pct >= -10 THEN 'Slight Decline'
          ELSE 'Significant Decline'
      END as growth_category
FROM growth_analysis
WHERE year_month >= ADD_MONTHS(SYSDATE, -24) -- Last 2 years
ORDER BY region, year_month;

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

Data state before query executes.

Output / Afterresult

After

EMP_NAME	DEPT_NAME
Alice Johnson	Engineering
Bob Smith	Marketing
Carol White	Engineering
Dave Brown	NULL

4 rows — all employees kept, NULL where no matching department

Key Takeaway

Use LAG window function or self-joins to compare current period values with previous period values and calculate growth percentages.

Q150. Find duplicate rows in a table without using GROUP BY.

Answer

Use window functions like ROW_NUMBER() or analytical functions to identify duplicate records without GROUP BY clause.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

CodeOracle SQL

SQL

-- Method 1: Using ROW_NUMBER() to find duplicates
SELECT employee_id, first_name, last_name, email
FROM (
    SELECT employee_id, first_name, last_name, email,
           ROW_NUMBER() OVER (PARTITION BY first_name, last_name, email ORDER BY employee_id) as rn
    FROM employees
)
WHERE rn > 1;

-- Method 2: Using COUNT() OVER() window function
SELECT employee_id, first_name, last_name, email, duplicate_count
FROM (
    SELECT employee_id, first_name, last_name, email,
           COUNT(*) OVER (PARTITION BY first_name, last_name, email) as duplicate_count
    FROM employees
)
WHERE duplicate_count > 1;

-- Method 3: Using EXISTS with correlation
SELECT DISTINCT e1.employee_id, e1.first_name, e1.last_name, e1.email
FROM employees e1
WHERE EXISTS (
    SELECT 1 FROM employees e2 
    WHERE e2.first_name = e1.first_name 
      AND e2.last_name = e1.last_name 
      AND e2.email = e1.email 
      AND e2.ROWID > e1.ROWID
);

-- Method 4: Using MINUS operation
SELECT * FROM employees
MINUS 
SELECT * FROM (
    SELECT DISTINCT * FROM employees
);

-- Method 5: Using analytical function DENSE_RANK
SELECT employee_id, first_name, last_name, email
FROM (
    SELECT employee_id, first_name, last_name, email,
           DENSE_RANK() OVER (PARTITION BY first_name, last_name, email ORDER BY employee_id) as rnk
    FROM employees
)
WHERE rnk > 1;

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

Data state before query executes.

Output / Afterresult

After

EMP_NAME	SALARY	ROW_NUM	RANK	DENSE_RANK
Dave Brown	70,000	1	1	1
Bob Smith	62,000	2	2	2
Carol White	55,000	3	3	3
Alice Johnson	50,000	4	4	4

All 4 employees ranked by salary DESC — no ties so all methods agree

Key Takeaway

Use window functions like ROW_NUMBER() or analytical functions to identify duplicate records without GROUP BY clause.

Q151. Write a SQL query to find the top 10% of earners in a table.

Answer

Use PERCENTILE_CONT, NTILE, or calculate percentage-based cutoffs to identify the top 10% salary earners.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

CodeOracle SQL

SQL

-- Method 1: Using PERCENTILE_CONT to find 90th percentile
WITH salary_percentile AS (
    SELECT PERCENTILE_CONT(0.9) WITHIN GROUP (ORDER BY salary) as top_10_percent_cutoff
    FROM employees
)
SELECT e.employee_id, e.first_name, e.last_name, e.salary
FROM employees e, salary_percentile sp
WHERE e.salary >= sp.top_10_percent_cutoff
ORDER BY e.salary DESC;

-- Method 2: Using NTILE to divide into 10 groups
SELECT employee_id, first_name, last_name, salary, salary_decile
FROM (
    SELECT employee_id, first_name, last_name, salary,
           NTILE(10) OVER (ORDER BY salary DESC) as salary_decile
    FROM employees
)
WHERE salary_decile = 1;

-- Method 3: Using PERCENT_RANK function
SELECT employee_id, first_name, last_name, salary, salary_percentile
FROM (
    SELECT employee_id, first_name, last_name, salary,
           PERCENT_RANK() OVER (ORDER BY salary DESC) as salary_percentile
    FROM employees
)
WHERE salary_percentile <= 0.1;

-- Method 4: Calculate top 10% using COUNT and ROWNUM
SELECT employee_id, first_name, last_name, salary
FROM (
    SELECT employee_id, first_name, last_name, salary,
           ROWNUM as rn,
           ROUND((SELECT COUNT(*) FROM employees) * 0.1) as top_10_percent_count
    FROM (
        SELECT employee_id, first_name, last_name, salary
        FROM employees
        ORDER BY salary DESC
    )
)
WHERE rn <= top_10_percent_count;

-- Method 5: Using analytical function with percentage calculation
WITH ranked_salaries AS (
    SELECT employee_id, first_name, last_name, salary,
           ROW_NUMBER() OVER (ORDER BY salary DESC) as salary_rank,
           COUNT(*) OVER () as total_employees
    FROM employees
)
SELECT employee_id, first_name, last_name, salary, 
       ROUND((salary_rank / total_employees) * 100, 2) as percentile
FROM ranked_salaries
WHERE salary_rank <= (total_employees * 0.1)
ORDER BY salary DESC;

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

Data state before query executes.

Output / Afterresult

After

EMP_NAME	SALARY	ROW_NUM	RANK	DENSE_RANK
Dave Brown	70,000	1	1	1
Bob Smith	62,000	2	2	2
Carol White	55,000	3	3	3
Alice Johnson	50,000	4	4	4

All 4 employees ranked by salary DESC — no ties so all methods agree

Key Takeaway

Use PERCENTILE_CONT, NTILE, or calculate percentage-based cutoffs to identify the top 10% salary earners.

Q152. Write a SQL query to find all employees who have never taken a leave.

Answer

Use NOT EXISTS, LEFT JOIN with NULL check, or NOT IN to find employees without any leave records.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

CodeOracle SQL

SQL

-- Assuming we have an employee_leaves table
-- CREATE TABLE employee_leaves (leave_id NUMBER, employee_id NUMBER, leave_date DATE, leave_type VARCHAR2(20));

-- Method 1: Using NOT EXISTS (most efficient)
SELECT e.employee_id, e.first_name, e.last_name, e.department_id
FROM employees e
WHERE NOT EXISTS (
    SELECT 1 
    FROM employee_leaves el 
    WHERE el.employee_id = e.employee_id
);

-- Method 2: Using LEFT JOIN with NULL check
SELECT e.employee_id, e.first_name, e.last_name, e.department_id
FROM employees e
LEFT JOIN employee_leaves el ON e.employee_id = el.employee_id
WHERE el.employee_id IS NULL;

-- Method 3: Using NOT IN (be careful with NULLs)
SELECT employee_id, first_name, last_name, department_id
FROM employees
WHERE employee_id NOT IN (
    SELECT employee_id 
    FROM employee_leaves 
    WHERE employee_id IS NOT NULL
);

-- Method 4: Using MINUS operation
SELECT employee_id, first_name, last_name, department_id
FROM employees
MINUS
SELECT e.employee_id, e.first_name, e.last_name, e.department_id
FROM employees e
JOIN employee_leaves el ON e.employee_id = el.employee_id;

-- Method 5: Using COUNT with HAVING
SELECT e.employee_id, e.first_name, e.last_name, e.department_id,
       COUNT(el.leave_id) as leave_count
FROM employees e
LEFT JOIN employee_leaves el ON e.employee_id = el.employee_id
GROUP BY e.employee_id, e.first_name, e.last_name, e.department_id
HAVING COUNT(el.leave_id) = 0;

-- Method 6: Find employees with no leaves in specific time period
SELECT e.employee_id, e.first_name, e.last_name
FROM employees e
WHERE NOT EXISTS (
    SELECT 1 
    FROM employee_leaves el 
    WHERE el.employee_id = e.employee_id
      AND el.leave_date >= DATE '2024-01-01'
      AND el.leave_date <= DATE '2024-12-31'
);

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

Data state before query executes.

Output / Afterresult

After

EMP_NAME	DEPT_NAME
Alice Johnson	Engineering
Bob Smith	Marketing
Carol White	Engineering
Dave Brown	NULL

4 rows — all employees kept, NULL where no matching department

Key Takeaway

Use NOT EXISTS, LEFT JOIN with NULL check, or NOT IN to find employees without any leave records.

Q153. Write a SQL query to find all departments with more than one employee.

Answer

Use subqueries, window functions, or EXISTS to find departments with multiple employees without using GROUP BY in the main query.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

CodeOracle SQL

SQL

-- Method 1: Using subquery with COUNT
SELECT DISTINCT d.department_id, d.department_name
FROM departments d
WHERE (
    SELECT COUNT(*) 
    FROM employees e 
    WHERE e.department_id = d.department_id
) > 1;

-- Method 2: Using EXISTS with correlated subquery
SELECT d.department_id, d.department_name
FROM departments d
WHERE EXISTS (
    SELECT e1.employee_id
    FROM employees e1
    WHERE e1.department_id = d.department_id
      AND EXISTS (
          SELECT 1 
          FROM employees e2 
          WHERE e2.department_id = e1.department_id 
            AND e2.employee_id != e1.employee_id
      )
);

-- Method 3: Using window function COUNT() OVER()
SELECT DISTINCT department_id, 
       (SELECT department_name FROM departments WHERE department_id = e.department_id) as department_name
FROM (
    SELECT department_id,
           COUNT(*) OVER (PARTITION BY department_id) as emp_count
    FROM employees
)
WHERE emp_count > 1;

-- Method 4: Using MINUS to exclude single-employee departments
SELECT department_id, department_name
FROM departments
MINUS
SELECT d.department_id, d.department_name
FROM departments d
WHERE (
    SELECT COUNT(*) 
    FROM employees e 
    WHERE e.department_id = d.department_id
) <= 1;

-- Method 5: Using self-join
SELECT DISTINCT d.department_id, d.department_name
FROM departments d
WHERE EXISTS (
    SELECT 1
    FROM employees e1, employees e2
    WHERE e1.department_id = d.department_id
      AND e2.department_id = d.department_id
      AND e1.employee_id != e2.employee_id
);

-- Method 6: Show department details with employee count
SELECT department_id, department_name, employee_count
FROM (
    SELECT d.department_id, d.department_name,
           (SELECT COUNT(*) FROM employees e WHERE e.department_id = d.department_id) as employee_count
    FROM departments d
)
WHERE employee_count > 1
ORDER BY employee_count DESC;

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

Data state before query executes.

Output / Afterresult

After

EMP_NAME	SALARY	ROW_NUM	RANK	DENSE_RANK
Dave Brown	70,000	1	1	1
Bob Smith	62,000	2	2	2
Carol White	55,000	3	3	3
Alice Johnson	50,000	4	4	4

All 4 employees ranked by salary DESC — no ties so all methods agree

Key Takeaway

Use subqueries, window functions, or EXISTS to find departments with multiple employees without using GROUP BY in the main query.

Q154. Find the maximum value of a column for each group without using GROUP BY.

Answer

Use window functions like MAX() OVER() with PARTITION BY to find maximum values per group without GROUP BY clause.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

CodeOracle SQL

SQL

-- Method 1: Using MAX() OVER() window function
SELECT DISTINCT department_id,
       MAX(salary) OVER (PARTITION BY department_id) as max_salary_in_dept
FROM employees
ORDER BY department_id;

-- Method 2: Using correlated subquery
SELECT DISTINCT e1.department_id,
       (SELECT MAX(e2.salary) 
        FROM employees e2 
        WHERE e2.department_id = e1.department_id) as max_salary
FROM employees e1
ORDER BY e1.department_id;

-- Method 3: Show employees with their department's maximum salary
SELECT employee_id, first_name, last_name, department_id, salary,
       MAX(salary) OVER (PARTITION BY department_id) as dept_max_salary,
       CASE WHEN salary = MAX(salary) OVER (PARTITION BY department_id) 
            THEN 'HIGHEST PAID IN DEPT' 
            ELSE 'NOT HIGHEST' 
       END as salary_status
FROM employees
ORDER BY department_id, salary DESC;

-- Method 4: Using FIRST_VALUE with window function
SELECT DISTINCT department_id,
       FIRST_VALUE(salary) OVER (
           PARTITION BY department_id 
           ORDER BY salary DESC 
           ROWS UNBOUNDED PRECEDING
       ) as max_salary
FROM employees;

-- Method 5: Find employees who earn the maximum in their department
SELECT employee_id, first_name, last_name, department_id, salary
FROM (
    SELECT employee_id, first_name, last_name, department_id, salary,
           MAX(salary) OVER (PARTITION BY department_id) as dept_max_salary
    FROM employees
)
WHERE salary = dept_max_salary;

-- Method 6: Multiple column maximums per group
SELECT DISTINCT department_id,
       MAX(salary) OVER (PARTITION BY department_id) as max_salary,
       MAX(hire_date) OVER (PARTITION BY department_id) as latest_hire_date,
       MIN(salary) OVER (PARTITION BY department_id) as min_salary,
       COUNT(*) OVER (PARTITION BY department_id) as employee_count
FROM employees
ORDER BY department_id;

-- Method 7: Using RANK() to identify maximum values
SELECT department_id, salary as max_salary
FROM (
    SELECT department_id, salary,
           RANK() OVER (PARTITION BY department_id ORDER BY salary DESC) as salary_rank
    FROM employees
)
WHERE salary_rank = 1;

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

Data state before query executes.

Output / Afterresult

After

EMP_NAME	SALARY	ROW_NUM	RANK	DENSE_RANK
Dave Brown	70,000	1	1	1
Bob Smith	62,000	2	2	2
Carol White	55,000	3	3	3
Alice Johnson	50,000	4	4	4

All 4 employees ranked by salary DESC — no ties so all methods agree

Key Takeaway

Use window functions like MAX() OVER() with PARTITION BY to find maximum values per group without GROUP BY clause.

Q155. Write a SQL query to find all employees who have taken more than 3 leaves in a month.

Answer

Use window functions, subqueries with date functions, or analytical queries to count leaves per employee per month.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

CodeOracle SQL

SQL

-- Assuming employee_leaves table with columns: employee_id, leave_date, leave_type

-- Method 1: Using window function COUNT() OVER()
SELECT employee_id, year_month, monthly_leave_count
FROM (
    SELECT el.employee_id,
           TO_CHAR(el.leave_date, 'YYYY-MM') as year_month,
           COUNT(*) OVER (
               PARTITION BY el.employee_id, TO_CHAR(el.leave_date, 'YYYY-MM')
           ) as monthly_leave_count
    FROM employee_leaves el
)
WHERE monthly_leave_count > 3
ORDER BY employee_id, year_month;

-- Method 2: Using subquery with date functions
SELECT DISTINCT e.employee_id, e.first_name, e.last_name, 
       leave_month, monthly_leaves
FROM employees e
JOIN (
    SELECT employee_id,
           TO_CHAR(leave_date, 'YYYY-MM') as leave_month,
           COUNT(*) as monthly_leaves
    FROM employee_leaves
    WHERE leave_date IS NOT NULL
    GROUP BY employee_id, TO_CHAR(leave_date, 'YYYY-MM')
    HAVING COUNT(*) > 3
) monthly_counts ON e.employee_id = monthly_counts.employee_id;

-- Method 3: Using EXTRACT functions for year/month
SELECT el.employee_id,
       EXTRACT(YEAR FROM el.leave_date) as leave_year,
       EXTRACT(MONTH FROM el.leave_date) as leave_month,
       COUNT(*) as leaves_in_month
FROM employee_leaves el
WHERE el.leave_date >= ADD_MONTHS(SYSDATE, -12) -- Last 12 months
GROUP BY el.employee_id, EXTRACT(YEAR FROM el.leave_date), EXTRACT(MONTH FROM el.leave_date)
HAVING COUNT(*) > 3
ORDER BY el.employee_id, leave_year, leave_month;

-- Method 4: Detailed analysis with employee information
SELECT e.employee_id, e.first_name, e.last_name, e.department_id,
       leave_analysis.leave_month,
       leave_analysis.leave_count,
       leave_analysis.leave_types
FROM employees e
JOIN (
    SELECT employee_id,
           TO_CHAR(leave_date, 'Mon YYYY') as leave_month,
           COUNT(*) as leave_count,
           LISTAGG(leave_type, ', ') WITHIN GROUP (ORDER BY leave_date) as leave_types
    FROM employee_leaves
    GROUP BY employee_id, TO_CHAR(leave_date, 'Mon YYYY'), TO_CHAR(leave_date, 'YYYY-MM')
    HAVING COUNT(*) > 3
) leave_analysis ON e.employee_id = leave_analysis.employee_id
ORDER BY e.employee_id, leave_analysis.leave_month;

-- Method 5: Using TRUNC function for month boundaries
WITH monthly_leaves AS (
    SELECT employee_id,
           TRUNC(leave_date, 'MM') as month_start,
           COUNT(*) as leave_count,
           MIN(leave_date) as first_leave,
           MAX(leave_date) as last_leave
    FROM employee_leaves
    GROUP BY employee_id, TRUNC(leave_date, 'MM')
    HAVING COUNT(*) > 3
)
SELECT e.employee_id, e.first_name, e.last_name,
       TO_CHAR(ml.month_start, 'Month YYYY') as leave_month,
       ml.leave_count,
       ml.first_leave,
       ml.last_leave
FROM employees e
JOIN monthly_leaves ml ON e.employee_id = ml.employee_id
ORDER BY e.employee_id, ml.month_start DESC;

-- Method 6: Current month analysis
SELECT e.employee_id, e.first_name, e.last_name,
       COUNT(el.leave_id) as current_month_leaves
FROM employees e
JOIN employee_leaves el ON e.employee_id = el.employee_id
WHERE TRUNC(el.leave_date, 'MM') = TRUNC(SYSDATE, 'MM')
GROUP BY e.employee_id, e.first_name, e.last_name
HAVING COUNT(el.leave_id) > 3;

-- Method 7: Trending analysis - employees with pattern of high leaves
SELECT employee_id,
       COUNT(CASE WHEN leave_count > 3 THEN 1 END) as months_with_high_leaves,
       AVG(leave_count) as avg_monthly_leaves,
       MAX(leave_count) as max_monthly_leaves
FROM (
    SELECT employee_id,
           TO_CHAR(leave_date, 'YYYY-MM') as month_year,
           COUNT(*) as leave_count
    FROM employee_leaves
    WHERE leave_date >= ADD_MONTHS(TRUNC(SYSDATE, 'MM'), -6)
    GROUP BY employee_id, TO_CHAR(leave_date, 'YYYY-MM')
) monthly_summary
GROUP BY employee_id
HAVING COUNT(CASE WHEN leave_count > 3 THEN 1 END) >= 2
ORDER BY months_with_high_leaves DESC;

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

Data state before query executes.

Output / Afterresult

After

EMP_NAME	DEPT_NAME
Alice Johnson	Engineering
Bob Smith	Marketing
Carol White	Engineering

3 rows — only rows with matching dept_id in both tables (Dave Brown excluded — no dept)

Key Takeaway

Use window functions, subqueries with date functions, or analytical queries to count leaves per employee per month.

Q156. How do NVL, NVL2, and COALESCE handle NULL values in Oracle?

Q157. How do you use TO_CHAR and TO_DATE to convert between dates and strings in Oracle?

Advanced Oracle: Hierarchical, Recursive & Pattern Queries

Q158. Find employees who have more direct reports (subordinates) than their manager in Oracle.

Answer

Use hierarchical queries with CONNECT BY and analytical functions to compare subordinate counts across management levels.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

CodeOracle SQL

SQL

-- Find employees with more direct reports than their manager
WITH employee_reports AS (
    SELECT 
        e.employee_id,
        e.first_name,
        e.last_name,
        e.manager_id,
        COUNT(subordinate.employee_id) as direct_reports
    FROM employees e
    LEFT JOIN employees subordinate ON e.employee_id = subordinate.manager_id
    GROUP BY e.employee_id, e.first_name, e.last_name, e.manager_id
),
manager_reports AS (
    SELECT 
        m.employee_id as manager_id,
        COUNT(emp.employee_id) as manager_direct_reports
    FROM employees m
    LEFT JOIN employees emp ON m.employee_id = emp.manager_id
    GROUP BY m.employee_id
)
SELECT 
    er.employee_id,
    er.first_name,
    er.last_name,
    er.direct_reports as employee_direct_reports,
    mr.manager_direct_reports,
    er.direct_reports - mr.manager_direct_reports as difference
FROM employee_reports er
JOIN manager_reports mr ON er.manager_id = mr.manager_id
WHERE er.direct_reports > mr.manager_direct_reports
ORDER BY difference DESC;

-- Alternative using hierarchical query
SELECT 
    emp.employee_id,
    emp.first_name || ' ' || emp.last_name as employee_name,
    emp_reports.direct_reports as employee_subordinates,
    mgr_reports.direct_reports as manager_subordinates
FROM employees emp
JOIN (
    SELECT manager_id, COUNT(*) as direct_reports
    FROM employees
    WHERE manager_id IS NOT NULL
    GROUP BY manager_id
) emp_reports ON emp.employee_id = emp_reports.manager_id
JOIN (
    SELECT manager_id, COUNT(*) as direct_reports
    FROM employees
    WHERE manager_id IS NOT NULL
    GROUP BY manager_id
) mgr_reports ON emp.manager_id = mgr_reports.manager_id
WHERE emp_reports.direct_reports > mgr_reports.direct_reports;

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

Data state before query executes.

Output / Afterresult

After

EMPLOYEE	MANAGER
Bob Smith	Alice Johnson
Carol White	Alice Johnson
Dave Brown	Bob Smith

3 rows — self join pairs each employee with their manager's name

Key Takeaway

Use hierarchical queries with CONNECT BY and analytical functions to compare subordinate counts across management levels.

Q159. How do you write recursive CTEs to handle hierarchical data in Oracle?

Answer

Recursive CTEs allow you to query hierarchical data by defining an anchor member (base case) and recursive member that references itself until a termination condition is met.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

CodeOracle SQL

SQL

-- Basic recursive CTE structure
WITH RECURSIVE hierarchy_cte AS (
   -- Anchor member (base case)
   SELECT employee_id, name, manager_id, 1 as level
   FROM employees 
   WHERE manager_id IS NULL
   
   UNION ALL
   
   -- Recursive member
   SELECT e.employee_id, e.name, e.manager_id, h.level + 1
   FROM employees e
   INNER JOIN hierarchy_cte h ON e.manager_id = h.employee_id
)
SELECT * FROM hierarchy_cte;

-- Employee hierarchy with path
WITH RECURSIVE emp_hierarchy AS (
   SELECT employee_id, name, manager_id, name as path, 1 as level
   FROM employees
   WHERE manager_id IS NULL
   
   UNION ALL
   
   SELECT e.employee_id, e.name, e.manager_id, 
          h.path || ' -> ' || e.name as path, h.level + 1
   FROM employees e
   JOIN emp_hierarchy h ON e.manager_id = h.employee_id
   WHERE h.level < 10 -- Prevent infinite recursion
)
SELECT employee_id, name, level, path
FROM emp_hierarchy
ORDER BY level, name;

-- Alternative Oracle CONNECT BY syntax
SELECT employee_id, name, manager_id, LEVEL,
      SYS_CONNECT_BY_PATH(name, ' -> ') as hierarchy_path
FROM employees
START WITH manager_id IS NULL
CONNECT BY PRIOR employee_id = manager_id
ORDER SIBLINGS BY name;

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

Data state before query executes.

Output / Afterresult

After

LVL	EMP_ID	EMP_NAME	MANAGER_ID
1	105	CEO	NULL
2	101	Alice Johnson	105
2	102	Bob Smith	105
3	103	Carol White	101

Recursive CTE/CONNECT BY traverses hierarchy — LEVEL = depth from root

Key Takeaway

Recursive CTEs allow you to query hierarchical data by defining an anchor member (base case) and recursive member that references itself until a termination condition is met.

Q160. Write a query to find gaps and islands in sequential data.

Answer

Use window functions and analytical techniques to identify missing sequences (gaps) and continuous sequences (islands) in data.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

CodeOracle SQL

SQL

-- Sample data: employee badge access logs
CREATE TABLE badge_access (
   employee_id NUMBER,
   access_date DATE
);

-- Finding gaps in sequential dates
WITH date_sequence AS (
   SELECT employee_id, access_date,
          LAG(access_date) OVER (PARTITION BY employee_id ORDER BY access_date) as prev_date,
          access_date - LAG(access_date) OVER (PARTITION BY employee_id ORDER BY access_date) as day_diff
   FROM badge_access
),
gaps AS (
   SELECT employee_id, prev_date + 1 as gap_start, access_date - 1 as gap_end
   FROM date_sequence
   WHERE day_diff > 1
)
SELECT employee_id, gap_start, gap_end, gap_end - gap_start + 1 as gap_days
FROM gaps;

-- Finding islands (consecutive sequences)
WITH numbered_sequence AS (
   SELECT employee_id, access_date,
          ROW_NUMBER() OVER (PARTITION BY employee_id ORDER BY access_date) as rn,
          access_date - ROW_NUMBER() OVER (PARTITION BY employee_id ORDER BY access_date) as island_group
   FROM badge_access
),
islands AS (
   SELECT employee_id, island_group, 
          MIN(access_date) as island_start,
          MAX(access_date) as island_end,
          COUNT(*) as consecutive_days
   FROM numbered_sequence
   GROUP BY employee_id, island_group
)
SELECT employee_id, island_start, island_end, consecutive_days
FROM islands
WHERE consecutive_days >= 5 -- Islands of 5+ consecutive days
ORDER BY employee_id, island_start;

-- Finding gaps in numeric sequences (like order numbers)
WITH order_gaps AS (
   SELECT order_id,
          LAG(order_id) OVER (ORDER BY order_id) as prev_order_id,
          order_id - LAG(order_id) OVER (ORDER BY order_id) - 1 as gap_size
   FROM orders
)
SELECT prev_order_id + 1 as missing_start,
      order_id - 1 as missing_end,
      gap_size as missing_count
FROM order_gaps
WHERE gap_size > 0;

-- Generate missing sequence numbers
WITH RECURSIVE missing_orders AS (
   SELECT prev_order_id + 1 as missing_id, order_id - 1 as max_missing
   FROM order_gaps
   WHERE gap_size > 0
   
   UNION ALL
   
   SELECT missing_id + 1, max_missing
   FROM missing_orders
   WHERE missing_id < max_missing
)
SELECT missing_id as missing_order_id
FROM missing_orders
ORDER BY missing_id;

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

Data state before query executes.

Output / Afterresult

After

EMP_NAME	SALARY	PREV_SALARY	NEXT_SALARY
Alice Johnson	50,000	NULL	55,000
Carol White	55,000	50,000	62,000
Bob Smith	62,000	55,000	70,000
Dave Brown	70,000	62,000	NULL

LAG = previous row value, LEAD = next row value in ordered result

Key Takeaway

Use window functions and analytical techniques to identify missing sequences (gaps) and continuous sequences (islands) in data.

Q161. Write a query to find overlapping date ranges in Oracle.

Answer

Use self-joins or window functions to identify records with overlapping start and end dates, commonly used for scheduling and booking systems.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

CodeOracle SQL

SQL

-- Sample table for bookings with date ranges
CREATE TABLE room_bookings (
   booking_id NUMBER,
   room_id NUMBER,
   guest_name VARCHAR2(100),
   start_date DATE,
   end_date DATE
);

-- Insert sample data
INSERT ALL
   INTO room_bookings VALUES (1, 101, 'John Doe', DATE '2024-01-10', DATE '2024-01-15')
   INTO room_bookings VALUES (2, 101, 'Jane Smith', DATE '2024-01-12', DATE '2024-01-18')
   INTO room_bookings VALUES (3, 101, 'Bob Wilson', DATE '2024-01-20', DATE '2024-01-25')
   INTO room_bookings VALUES (4, 102, 'Alice Brown', DATE '2024-01-10', DATE '2024-01-15')
   INTO room_bookings VALUES (5, 101, 'Tom Green', DATE '2024-01-14', DATE '2024-01-16')
SELECT * FROM dual;

-- Method 1: Self-join to find overlapping bookings
SELECT b1.booking_id as booking1_id,
      b1.guest_name as guest1,
      b1.start_date as start1,
      b1.end_date as end1,
      b2.booking_id as booking2_id,
      b2.guest_name as guest2,
      b2.start_date as start2,
      b2.end_date as end2,
      -- Calculate overlap period
      GREATEST(b1.start_date, b2.start_date) as overlap_start,
      LEAST(b1.end_date, b2.end_date) as overlap_end,
      LEAST(b1.end_date, b2.end_date) - GREATEST(b1.start_date, b2.start_date) + 1 as overlap_days
FROM room_bookings b1
JOIN room_bookings b2 ON b1.room_id = b2.room_id
                     AND b1.booking_id < b2.booking_id
WHERE b1.start_date <= b2.end_date
 AND b1.end_date >= b2.start_date;

-- Method 2: Using window functions to detect overlaps
WITH booking_analysis AS (
   SELECT booking_id, room_id, guest_name, start_date, end_date,
          LAG(end_date) OVER (PARTITION BY room_id ORDER BY start_date) as prev_end_date,
          LAG(guest_name) OVER (PARTITION BY room_id ORDER BY start_date) as prev_guest,
          LAG(booking_id) OVER (PARTITION BY room_id ORDER BY start_date) as prev_booking_id
   FROM room_bookings
)
SELECT booking_id, room_id, guest_name, start_date, end_date,
      prev_booking_id, prev_guest, prev_end_date,
      'OVERLAP DETECTED' as conflict_status
FROM booking_analysis
WHERE start_date <= prev_end_date;

-- Method 3: Find all conflicts for a specific room
SELECT DISTINCT b1.booking_id, b1.guest_name, b1.start_date, b1.end_date
FROM room_bookings b1
WHERE b1.room_id = 101
 AND EXISTS (
     SELECT 1 FROM room_bookings b2
     WHERE b2.room_id = b1.room_id
       AND b2.booking_id != b1.booking_id
       AND b2.start_date <= b1.end_date
       AND b2.end_date >= b1.start_date
 )
ORDER BY b1.start_date;

-- Method 4: Advanced overlap analysis with conflict resolution
WITH overlapping_groups AS (
   SELECT booking_id, room_id, guest_name, start_date, end_date,
          -- Create groups of overlapping bookings
          SUM(CASE WHEN start_date <= LAG(end_date) OVER (PARTITION BY room_id ORDER BY start_date) 
                   THEN 0 ELSE 1 END) 
          OVER (PARTITION BY room_id ORDER BY start_date 
                ROWS UNBOUNDED PRECEDING) as group_id
   FROM room_bookings
),
conflict_summary AS (
   SELECT room_id, group_id,
          COUNT(*) as conflicting_bookings,
          MIN(start_date) as group_start,
          MAX(end_date) as group_end,
          LISTAGG(guest_name, ', ') WITHIN GROUP (ORDER BY start_date) as affected_guests
   FROM overlapping_groups
   GROUP BY room_id, group_id
   HAVING COUNT(*) > 1
)
SELECT room_id, 
      'Room ' || room_id || ' has ' || conflicting_bookings || ' overlapping bookings' as conflict_description,
      group_start, group_end,
      affected_guests,
      group_end - group_start + 1 as total_conflict_days
FROM conflict_summary;

-- Method 5: Check for gaps between bookings (opposite of overlaps)
WITH booking_gaps AS (
   SELECT room_id, booking_id, guest_name, start_date, end_date,
          LAG(end_date) OVER (PARTITION BY room_id ORDER BY start_date) as prev_end,
          start_date - LAG(end_date) OVER (PARTITION BY room_id ORDER BY start_date) - 1 as gap_days
   FROM room_bookings
)
SELECT room_id, booking_id, guest_name,
      prev_end + 1 as gap_start,
      start_date - 1 as gap_end,
      gap_days
FROM booking_gaps
WHERE gap_days > 0;

-- Method 6: Prevent overlaps with validation function
CREATE OR REPLACE FUNCTION check_booking_overlap(
   p_room_id NUMBER,
   p_start_date DATE,
   p_end_date DATE,
   p_exclude_booking_id NUMBER DEFAULT NULL
) RETURN VARCHAR2 IS
   v_overlap_count NUMBER;
   v_conflicting_guest VARCHAR2(100);
BEGIN
   SELECT COUNT(*), MAX(guest_name)
   INTO v_overlap_count, v_conflicting_guest
   FROM room_bookings
   WHERE room_id = p_room_id
     AND (booking_id != p_exclude_booking_id OR p_exclude_booking_id IS NULL)
     AND start_date <= p_end_date
     AND end_date >= p_start_date;
   
   IF v_overlap_count > 0 THEN
       RETURN 'CONFLICT: Overlaps with booking for ' || v_conflicting_guest;
   ELSE
       RETURN 'OK: No conflicts detected';
   END IF;
END;
/

-- Test the validation function
SELECT check_booking_overlap(101, DATE '2024-01-13', DATE '2024-01-17') as validation_result
FROM dual;

-- Method 7: Timeline view of bookings
SELECT room_id,
      start_date,
      end_date,
      guest_name,
      -- Show booking duration
      end_date - start_date + 1 as booking_days,
      -- Show overlapping periods
      CASE 
          WHEN start_date <= LAG(end_date) OVER (PARTITION BY room_id ORDER BY start_date)
          THEN 'OVERLAPS WITH PREVIOUS'
          ELSE 'NO CONFLICT'
      END as overlap_status
FROM room_bookings
ORDER BY room_id, start_date;

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

Data state before query executes.

Output / Afterresult

After

EMP_NAME	DEPT_NAME
Alice Johnson	Engineering
Bob Smith	Marketing
Carol White	Engineering

3 rows — only rows with matching dept_id in both tables (Dave Brown excluded — no dept)

Key Takeaway

Use self-joins or window functions to identify records with overlapping start and end dates, commonly used for scheduling and booking systems.

Q162. How do CONNECT BY and START WITH work for hierarchical queries in Oracle?

Oracle Performance & Partitioning

Q163. Explain Oracle execution plans and how to optimize slow queries.

Answer

Execution plans show how Oracle processes SQL statements.

Use EXPLAIN PLAN, AUTOTRACE, and SQL tuning techniques to optimize performance.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

CodeOracle SQL

SQL

-- Generate and view execution plan
EXPLAIN PLAN FOR
SELECT e.first_name, d.department_name
FROM employees e
JOIN departments d ON e.department_id = d.department_id
WHERE e.salary > 50000;

-- View the execution plan
SELECT * FROM TABLE(DBMS_XPLAN.DISPLAY);

-- Alternative: Use AUTOTRACE
SET AUTOTRACE ON EXPLAIN;
SELECT COUNT(*) FROM employees WHERE salary > 50000;
SET AUTOTRACE OFF;

-- Analyze execution plan components
SELECT operation, options, object_name, cost, cardinality, bytes
FROM plan_table
WHERE plan_id = (SELECT MAX(plan_id) FROM plan_table);

-- Common optimization techniques
-- 1. Add appropriate indexes
CREATE INDEX idx_emp_salary ON employees(salary);
CREATE INDEX idx_emp_dept_join ON employees(department_id, salary);

-- 2. Use hints when necessary (use sparingly)
SELECT /*+ USE_INDEX(e, idx_emp_salary) */ 
      e.first_name, e.salary
FROM employees e
WHERE e.salary > 50000;

-- 3. Rewrite inefficient queries
-- Bad: Using functions in WHERE clause
SELECT * FROM employees WHERE UPPER(first_name) = 'JOHN';

-- Good: Use function-based index or avoid function
CREATE INDEX idx_emp_upper_name ON employees(UPPER(first_name));
-- or
SELECT * FROM employees WHERE first_name = 'John';

-- 4. Optimize JOIN order and conditions
-- Ensure proper JOIN conditions and consider driving table
SELECT /*+ LEADING(d) USE_NL(e) */
      e.first_name, d.department_name
FROM departments d
JOIN employees e ON d.department_id = e.department_id
WHERE d.department_name = 'Sales';

-- 5. Use EXISTS instead of IN for better performance
-- Less efficient
SELECT * FROM departments 
WHERE department_id IN (SELECT department_id FROM employees);

-- More efficient  
SELECT * FROM departments d
WHERE EXISTS (SELECT 1 FROM employees e WHERE e.department_id = d.department_id);

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

Data state before query executes.

Output / Afterresult

After

EMP_NAME	DEPT_NAME
Alice Johnson	Engineering
Bob Smith	Marketing
Carol White	Engineering

3 rows — only rows with matching dept_id in both tables (Dave Brown excluded — no dept)

Key Takeaway

Execution plans show how Oracle processes SQL statements.

Q164. How do you handle large datasets and implement pagination efficiently in Oracle?

Answer

Use ROWNUM, ROW_NUMBER(), or OFFSET/FETCH for pagination.

Consider performance implications and use appropriate indexing strategies.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

CodeOracle SQL

SQL

-- Method 1: OFFSET/FETCH (Oracle 12c+) - Recommended
SELECT employee_id, first_name, last_name, salary
FROM employees
ORDER BY salary DESC
OFFSET 20 ROWS FETCH NEXT 10 ROWS ONLY;

-- Method 2: ROW_NUMBER() with subquery
SELECT employee_id, first_name, last_name, salary
FROM (
   SELECT employee_id, first_name, last_name, salary,
          ROW_NUMBER() OVER (ORDER BY salary DESC) as rn
   FROM employees
)
WHERE rn BETWEEN 21 AND 30;

-- Method 3: ROWNUM (Oracle 11g and earlier)
SELECT employee_id, first_name, last_name, salary
FROM (
   SELECT employee_id, first_name, last_name, salary, ROWNUM rn
   FROM (
       SELECT employee_id, first_name, last_name, salary
       FROM employees
       ORDER BY salary DESC
   )
   WHERE ROWNUM <= 30
)
WHERE rn >= 21;

-- Efficient pagination with seek method (for large datasets)
-- First page
SELECT employee_id, first_name, last_name, salary
FROM employees
WHERE salary IS NOT NULL
ORDER BY salary DESC, employee_id
FETCH NEXT 10 ROWS ONLY;

-- Subsequent pages (using last row from previous page)
SELECT employee_id, first_name, last_name, salary
FROM employees
WHERE (salary < :last_salary OR (salary = :last_salary AND employee_id > :last_employee_id))
 AND salary IS NOT NULL
ORDER BY salary DESC, employee_id
FETCH NEXT 10 ROWS ONLY;

-- Performance optimization for pagination
-- Create appropriate indexes
CREATE INDEX idx_emp_salary_id ON employees(salary DESC, employee_id);

-- Count total rows efficiently (for pagination info)
SELECT COUNT(*) OVER() as total_count,
      employee_id, first_name, salary
FROM employees
WHERE salary > 50000
ORDER BY salary DESC
OFFSET 0 ROWS FETCH NEXT 10 ROWS ONLY;

-- Batch processing for large datasets
DECLARE
   CURSOR emp_cursor IS
       SELECT employee_id, salary
       FROM employees
       ORDER BY employee_id;
   
   TYPE emp_array IS TABLE OF emp_cursor%ROWTYPE;
   emp_batch emp_array;
   
   batch_size CONSTANT NUMBER := 1000;
BEGIN
   OPEN emp_cursor;
   LOOP
       FETCH emp_cursor BULK COLLECT INTO emp_batch LIMIT batch_size;
       
       -- Process batch
       FOR i IN 1..emp_batch.COUNT LOOP
           -- Update salary logic here
           UPDATE employees 
           SET salary = emp_batch(i).salary * 1.1 
           WHERE employee_id = emp_batch(i).employee_id;
       END LOOP;
       
       COMMIT; -- Commit each batch
       
       EXIT WHEN emp_cursor%NOTFOUND;
   END LOOP;
   CLOSE emp_cursor;
END;
/

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

Data state before query executes.

Output / Afterresult

After

OPERATION	ROWS	CONTEXT_SWITCHES
Row-by-row	4	4 (slow)
BULK COLLECT	4	1 (fast)

BULK COLLECT fetches all rows at once — reduces SQL↔PL/SQL round trips

Key Takeaway

Use ROWNUM, ROW_NUMBER(), or OFFSET/FETCH for pagination.

Q165. Explain Oracle partitioning and write queries for partitioned tables.

Answer

Oracle partitioning divides large tables into smaller, manageable pieces based on partition keys, improving performance and maintenance.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

CodeOracle SQL

SQL

-- Range partitioning by date
CREATE TABLE sales_data (
   sale_id NUMBER,
   sale_date DATE,
   customer_id NUMBER,
   amount NUMBER
)
PARTITION BY RANGE (sale_date) (
   PARTITION p_2022 VALUES LESS THAN (DATE '2023-01-01'),
   PARTITION p_2023 VALUES LESS THAN (DATE '2024-01-01'),
   PARTITION p_2024 VALUES LESS THAN (DATE '2025-01-01'),
   PARTITION p_future VALUES LESS THAN (MAXVALUE)
);

-- List partitioning by region
CREATE TABLE customer_regional (
   customer_id NUMBER,
   name VARCHAR2(100),
   region VARCHAR2(50),
   signup_date DATE
)
PARTITION BY LIST (region) (
   PARTITION p_north VALUES ('North', 'Northeast'),
   PARTITION p_south VALUES ('South', 'Southeast'),
   PARTITION p_west VALUES ('West', 'Northwest'),
   PARTITION p_other VALUES (DEFAULT)
);

-- Hash partitioning for even distribution
CREATE TABLE user_sessions (
   session_id VARCHAR2(50),
   user_id NUMBER,
   start_time TIMESTAMP,
   end_time TIMESTAMP
)
PARTITION BY HASH (user_id) PARTITIONS 8;

-- Composite partitioning (Range-Hash)
CREATE TABLE order_details (
   order_id NUMBER,
   order_date DATE,
   customer_id NUMBER,
   product_id NUMBER,
   quantity NUMBER
)
PARTITION BY RANGE (order_date)
SUBPARTITION BY HASH (customer_id) SUBPARTITIONS 4 (
   PARTITION p_2023_q1 VALUES LESS THAN (DATE '2023-04-01'),
   PARTITION p_2023_q2 VALUES LESS THAN (DATE '2023-07-01'),
   PARTITION p_2023_q3 VALUES LESS THAN (DATE '2023-10-01'),
   PARTITION p_2023_q4 VALUES LESS THAN (DATE '2024-01-01')
);

-- Query specific partitions
SELECT * FROM sales_data PARTITION (p_2023)
WHERE amount > 1000;

-- Cross-partition queries
SELECT EXTRACT(YEAR FROM sale_date) as year,
      SUM(amount) as total_sales
FROM sales_data
WHERE sale_date BETWEEN DATE '2022-01-01' AND DATE '2024-12-31'
GROUP BY EXTRACT(YEAR FROM sale_date);

-- Partition maintenance operations
-- Add new partition
ALTER TABLE sales_data ADD PARTITION p_2025 VALUES LESS THAN (DATE '2026-01-01');

-- Drop old partition
ALTER TABLE sales_data DROP PARTITION p_2022;

-- Split partition
ALTER TABLE sales_data SPLIT PARTITION p_future 
AT (DATE '2026-01-01') INTO (PARTITION p_2025, PARTITION p_future);

-- Partition-wise joins (both tables partitioned on same key)
SELECT s.sale_date, s.amount, c.name
FROM sales_data s
JOIN customer_regional c ON s.customer_id = c.customer_id
WHERE s.sale_date >= DATE '2023-01-01';

-- Check partition information
SELECT table_name, partition_name, high_value, num_rows
FROM user_tab_partitions
WHERE table_name = 'SALES_DATA'
ORDER BY partition_position;

-- Partition pruning examples
-- This query will only scan p_2023 partition
SELECT COUNT(*) FROM sales_data 
WHERE sale_date BETWEEN DATE '2023-06-01' AND DATE '2023-06-30';

-- Parallel operations on partitions
SELECT /*+ PARALLEL(s, 4) */ 
      EXTRACT(MONTH FROM sale_date) as month,
      SUM(amount) as monthly_sales
FROM sales_data s
WHERE EXTRACT(YEAR FROM sale_date) = 2023
GROUP BY EXTRACT(MONTH FROM sale_date);

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

Data state before query executes.

Output / Afterresult

After

EMP_NAME	DEPT_NAME
Alice Johnson	Engineering
Bob Smith	Marketing
Carol White	Engineering

3 rows — only rows with matching dept_id in both tables (Dave Brown excluded — no dept)

Key Takeaway

Oracle partitioning divides large tables into smaller, manageable pieces based on partition keys, improving performance and maintenance.

Q166. How do you optimize queries involving multiple large table joins?

Answer

Use proper indexing strategies, join order optimization, hints, and query rewriting techniques to improve performance of complex joins.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

CodeOracle SQL

SQL

-- Problem: Slow query with multiple large table joins
-- Original inefficient query
SELECT o.order_id, c.customer_name, p.product_name, od.quantity, od.price
FROM orders o
JOIN customers c ON o.customer_id = c.customer_id
JOIN order_details od ON o.order_id = od.order_id
JOIN products p ON od.product_id = p.product_id
WHERE o.order_date BETWEEN DATE '2023-01-01' AND DATE '2023-12-31'
 AND c.country = 'USA'
 AND p.category = 'Electronics';

-- Solution 1: Create appropriate indexes
CREATE INDEX idx_orders_date_customer ON orders(order_date, customer_id);
CREATE INDEX idx_customers_country ON customers(country, customer_id);
CREATE INDEX idx_products_category ON products(category, product_id);
CREATE INDEX idx_order_details_composite ON order_details(order_id, product_id);

-- Solution 2: Optimized query with hints and better structure
SELECT /*+ LEADING(c) USE_NL(o) USE_HASH(od p) */
      o.order_id, c.customer_name, p.product_name, od.quantity, od.price
FROM customers c
JOIN orders o ON c.customer_id = o.customer_id
JOIN order_details od ON o.order_id = od.order_id
JOIN products p ON od.product_id = p.product_id
WHERE c.country = 'USA'
 AND o.order_date BETWEEN DATE '2023-01-01' AND DATE '2023-12-31'
 AND p.category = 'Electronics';

-- Solution 3: Break into smaller queries with CTEs
WITH filtered_customers AS (
   SELECT customer_id, customer_name
   FROM customers
   WHERE country = 'USA'
),
filtered_orders AS (
   SELECT o.order_id, o.customer_id
   FROM orders o
   JOIN filtered_customers c ON o.customer_id = c.customer_id
   WHERE o.order_date BETWEEN DATE '2023-01-01' AND DATE '2023-12-31'
),
filtered_products AS (
   SELECT product_id, product_name
   FROM products
   WHERE category = 'Electronics'
)
SELECT fo.order_id, fc.customer_name, fp.product_name, od.quantity, od.price
FROM filtered_orders fo
JOIN filtered_customers fc ON fo.customer_id = fc.customer_id
JOIN order_details od ON fo.order_id = od.order_id
JOIN filtered_products fp ON od.product_id = fp.product_id;

-- Solution 4: Use EXISTS instead of JOINs when appropriate
SELECT o.order_id, 
      (SELECT customer_name FROM customers WHERE customer_id = o.customer_id) as customer_name
FROM orders o
WHERE o.order_date BETWEEN DATE '2023-01-01' AND DATE '2023-12-31'
 AND EXISTS (SELECT 1 FROM customers c WHERE c.customer_id = o.customer_id AND c.country = 'USA')
 AND EXISTS (
     SELECT 1 FROM order_details od 
     JOIN products p ON od.product_id = p.product_id
     WHERE od.order_id = o.order_id AND p.category = 'Electronics'
 );

-- Solution 5: Materialized views for frequently joined data
CREATE MATERIALIZED VIEW mv_customer_orders
BUILD IMMEDIATE
REFRESH FAST ON COMMIT
AS
SELECT o.order_id, o.order_date, o.customer_id, 
      c.customer_name, c.country, c.city
FROM orders o
JOIN customers c ON o.customer_id = c.customer_id;

-- Create indexes on materialized view
CREATE INDEX idx_mv_country_date ON mv_customer_orders(country, order_date);

-- Use materialized view in queries
SELECT mv.order_id, mv.customer_name, p.product_name, od.quantity
FROM mv_customer_orders mv
JOIN order_details od ON mv.order_id = od.order_id
JOIN products p ON od.product_id = p.product_id
WHERE mv.country = 'USA'
 AND mv.order_date BETWEEN DATE '2023-01-01' AND DATE '2023-12-31'
 AND p.category = 'Electronics';

-- Solution 6: Parallel processing for large datasets
SELECT /*+ PARALLEL(o, 4) PARALLEL(c, 4) PARALLEL(od, 4) PARALLEL(p, 4) */
      o.order_id, c.customer_name, p.product_name, od.quantity
FROM orders o
JOIN customers c ON o.customer_id = c.customer_id
JOIN order_details od ON o.order_id = od.order_id
JOIN products p ON od.product_id = p.product_id
WHERE o.order_date BETWEEN DATE '2023-01-01' AND DATE '2023-12-31';

-- Solution 7: Partitioning large tables
-- Partition orders by year
ALTER TABLE orders MODIFY
PARTITION BY RANGE (order_date) (
   PARTITION p_2022 VALUES LESS THAN (DATE '2023-01-01'),
   PARTITION p_2023 VALUES LESS THAN (DATE '2024-01-01'),
   PARTITION p_2024 VALUES LESS THAN (DATE '2025-01-01')
);

-- Solution 8: Query rewriting with aggregation pushdown
-- Instead of joining then aggregating
SELECT c.country, SUM(od.quantity * od.price) as total_sales
FROM (
   -- Pre-aggregate at order level
   SELECT o.customer_id, SUM(od.quantity * od.price) as order_total
   FROM orders o
   JOIN order_details od ON o.order_id = od.order_id
   WHERE o.order_date BETWEEN DATE '2023-01-01' AND DATE '2023-12-31'
   GROUP BY o.customer_id
) order_totals
JOIN customers c ON order_totals.customer_id = c.customer_id
GROUP BY c.country;

-- Performance monitoring query
SELECT sql_id, child_number, executions, elapsed_time/executions as avg_elapsed,
      buffer_gets/executions as avg_buffer_gets,
      sql_text
FROM v$sql
WHERE sql_text LIKE '%orders%customers%'
 AND executions > 0
ORDER BY elapsed_time/executions DESC;

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

Data state before query executes.

Output / Afterresult

After

EMP_NAME	DEPT_NAME
Alice Johnson	Engineering
Bob Smith	Marketing
Carol White	Engineering

3 rows — only rows with matching dept_id in both tables (Dave Brown excluded — no dept)

Key Takeaway

Use proper indexing strategies, join order optimization, hints, and query rewriting techniques to improve performance of complex joins.

Q167. How do you implement database connection pooling and session management in Oracle?

Answer

Use Oracle connection pooling features, DRCP (Database Resident Connection Pooling), and proper session management techniques for optimal performance.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

CodeOracle SQL

SQL

-- Enable Database Resident Connection Pooling (DRCP)
-- Run as SYSDBA
EXECUTE DBMS_CONNECTION_POOL.START_POOL();

-- Configure pool parameters
EXECUTE DBMS_CONNECTION_POOL.CONFIGURE_POOL(
   pool_name => 'SYS_DEFAULT_CONNECTION_POOL',
   minpool_size => 4,
   maxpool_size => 40,
   incrpool_size => 2,
   session_cached_cursors => 20,
   inactivity_timeout => 300,
   max_think_time => 600,
   max_use_session => 500000,
   max_lifetime_session => 86400
);

-- Check pool status
SELECT connection_pool, status, minsize, maxsize, poolsize, num_requests
FROM v$cpool_info;

-- Monitor pool statistics
SELECT * FROM v$cpool_stats;

-- Session management queries
-- Check current sessions
SELECT s.sid, s.serial#, s.username, s.status, s.program, s.machine,
      s.logon_time, s.last_call_et
FROM v$session s
WHERE s.username IS NOT NULL
ORDER BY s.last_call_et DESC;

-- Session resource usage
SELECT s.sid, s.username, s.program,
      st.value as cpu_used,
      sm.value as memory_used,
      si.value as logical_reads
FROM v$session s
LEFT JOIN v$sesstat st ON s.sid = st.sid AND st.statistic# = (SELECT statistic# FROM v$statname WHERE name = 'CPU used by this session')
LEFT JOIN v$sesstat sm ON s.sid = sm.sid AND sm.statistic# = (SELECT statistic# FROM v$statname WHERE name = 'session memory')
LEFT JOIN v$sesstat si ON s.sid = si.sid AND si.statistic# = (SELECT statistic# FROM v$statname WHERE name = 'session logical reads')
WHERE s.username IS NOT NULL;

-- Kill idle sessions
CREATE OR REPLACE PROCEDURE kill_idle_sessions(
   p_idle_minutes NUMBER DEFAULT 60,
   p_exclude_users VARCHAR2 DEFAULT 'SYS,SYSTEM'
) IS
   CURSOR idle_sessions_cur IS
       SELECT sid, serial#, username
       FROM v$session
       WHERE status = 'INACTIVE'
         AND last_call_et > (p_idle_minutes * 60)
         AND username IS NOT NULL
         AND username NOT IN (SELECT TRIM(REGEXP_SUBSTR(p_exclude_users, '[^,]+', 1, LEVEL))
                             FROM dual
                             CONNECT BY LEVEL <= REGEXP_COUNT(p_exclude_users, ',') + 1);
   
   v_sql VARCHAR2(200);
BEGIN
   FOR rec IN idle_sessions_cur LOOP
       v_sql := 'ALTER SYSTEM KILL SESSION ''' || rec.sid || ',' || rec.serial# || '''';
       EXECUTE IMMEDIATE v_sql;
       DBMS_OUTPUT.PUT_LINE('Killed session: ' || rec.username || ' (SID=' || rec.sid || ')');
   END LOOP;
END;
/

-- Connection pool monitoring procedure
CREATE OR REPLACE PROCEDURE monitor_connection_pool IS
   v_active_sessions NUMBER;
   v_inactive_sessions NUMBER;
   v_total_sessions NUMBER;
   v_cpu_usage NUMBER;
BEGIN
   -- Get session counts
   SELECT COUNT(CASE WHEN status = 'ACTIVE' THEN 1 END),
          COUNT(CASE WHEN status = 'INACTIVE' THEN 1 END),
          COUNT(*)
   INTO v_active_sessions, v_inactive_sessions, v_total_sessions
   FROM v$session
   WHERE username IS NOT NULL;
   
   -- Get CPU usage
   SELECT value INTO v_cpu_usage
   FROM v$sysstat
   WHERE name = 'CPU used by this session';
   
   -- Log statistics
   INSERT INTO connection_pool_stats (
       timestamp, active_sessions, inactive_sessions, 
       total_sessions, cpu_usage
   ) VALUES (
       SYSTIMESTAMP, v_active_sessions, v_inactive_sessions,
       v_total_sessions, v_cpu_usage
   );
   
   -- Alert if thresholds exceeded
   IF v_total_sessions > 100 THEN
       DBMS_OUTPUT.PUT_LINE('WARNING: High session count: ' || v_total_sessions);
   END IF;
   
   COMMIT;
END;
/

-- Create monitoring table
CREATE TABLE connection_pool_stats (
   timestamp TIMESTAMP,
   active_sessions NUMBER,
   inactive_sessions NUMBER,
   total_sessions NUMBER,
   cpu_usage NUMBER
);

-- Application context for session management
CREATE OR REPLACE CONTEXT app_context USING app_context_pkg;

CREATE OR REPLACE PACKAGE app_context_pkg IS
   PROCEDURE set_session_info(
       p_user_id VARCHAR2,
       p_app_name VARCHAR2,
       p_module VARCHAR2
   );
   PROCEDURE clear_session_info;
END;
/

CREATE OR REPLACE PACKAGE BODY app_context_pkg IS
   PROCEDURE set_session_info(
       p_user_id VARCHAR2,
       p_app_name VARCHAR2,
       p_module VARCHAR2
   ) IS
   BEGIN
       DBMS_SESSION.SET_CONTEXT('APP_CONTEXT', 'USER_ID', p_user_id);
       DBMS_SESSION.SET_CONTEXT('APP_CONTEXT', 'APP_NAME', p_app_name);
       DBMS_SESSION.SET_CONTEXT('APP_CONTEXT', 'MODULE', p_module);
       DBMS_APPLICATION_INFO.SET_MODULE(p_app_name, p_module);
   END;
   
   PROCEDURE clear_session_info IS
   BEGIN
       DBMS_SESSION.CLEAR_CONTEXT('APP_CONTEXT');
       DBMS_APPLICATION_INFO.SET_MODULE(NULL, NULL);
   END;
END;
/

-- Session pooling configuration for applications
-- JDBC connection pooling example configuration
/*
-- HikariCP configuration
maximumPoolSize=20
minimumIdle=5
connectionTimeout=30000
idleTimeout=600000
maxLifetime=1800000
leakDetectionThreshold=60000

-- Oracle UCP configuration
oracle.ucp.ConnectionPoolName=MyPool
oracle.ucp.InitialPoolSize=5
oracle.ucp.MinPoolSize=5
oracle.ucp.MaxPoolSize=20
oracle.ucp.ConnectionWaitTimeout=3
oracle.ucp.InactiveConnectionTimeout=30
oracle.ucp.AbandonedConnectionTimeout=30
*/

-- Session resource limits
ALTER PROFILE DEFAULT LIMIT
   SESSIONS_PER_USER 10
   CONNECT_TIME 480
   IDLE_TIME 30
   CPU_PER_SESSION UNLIMITED
   LOGICAL_READS_PER_SESSION UNLIMITED;

-- Monitor session waits
SELECT s.sid, s.username, s.status, w.event, w.wait_time, w.seconds_in_wait
FROM v$session s
LEFT JOIN v$session_wait w ON s.sid = w.sid
WHERE s.username IS NOT NULL
 AND w.event NOT LIKE '%message%'
ORDER BY w.seconds_in_wait DESC;

-- Session cleanup procedure
CREATE OR REPLACE PROCEDURE cleanup_sessions IS
BEGIN
   -- Kill sessions that have been inactive for more than 2 hours
   FOR rec IN (
       SELECT sid, serial#, username
       FROM v$session
       WHERE status = 'INACTIVE'
         AND last_call_et > 7200
         AND username NOT IN ('SYS', 'SYSTEM', 'DBSNMP')
   ) LOOP
       EXECUTE IMMEDIATE 'ALTER SYSTEM KILL SESSION ''' || rec.sid || ',' || rec.serial# || '''';
   END LOOP;
   
   -- Log cleanup activity
   INSERT INTO session_cleanup_log (cleanup_date, sessions_killed)
   SELECT SYSDATE, SQL%ROWCOUNT FROM dual;
   
   COMMIT;
END;
/

-- Create cleanup log table
CREATE TABLE session_cleanup_log (
   cleanup_date DATE,
   sessions_killed NUMBER
);

-- Schedule session cleanup job
BEGIN
   DBMS_SCHEDULER.CREATE_JOB(
       job_name => 'SESSION_CLEANUP_JOB',
       job_type => 'PLSQL_BLOCK',
       job_action => 'BEGIN cleanup_sessions; END;',
       start_date => SYSTIMESTAMP,
       repeat_interval => 'FREQ=HOURLY; INTERVAL=2',
       enabled => TRUE,
       comments => 'Cleanup idle sessions every 2 hours'
   );
END;
/

-- Connection pool health check
CREATE OR REPLACE FUNCTION connection_pool_health_check RETURN VARCHAR2 IS
   v_total_sessions NUMBER;
   v_active_sessions NUMBER;
   v_blocked_sessions NUMBER;
   v_health_status VARCHAR2(100);
BEGIN
   SELECT COUNT(*) INTO v_total_sessions
   FROM v$session WHERE username IS NOT NULL;
   
   SELECT COUNT(*) INTO v_active_sessions
   FROM v$session WHERE status = 'ACTIVE' AND username IS NOT NULL;
   
   SELECT COUNT(*) INTO v_blocked_sessions
   FROM v$session WHERE blocking_session IS NOT NULL;
   
   IF v_blocked_sessions > 5 THEN
       v_health_status := 'CRITICAL: ' || v_blocked_sessions || ' blocked sessions';
   ELSIF v_total_sessions > 80 THEN
       v_health_status := 'WARNING: High session count ' || v_total_sessions;
   ELSIF v_active_sessions < 5 THEN
       v_health_status := 'INFO: Low activity ' || v_active_sessions || ' active';
   ELSE
       v_health_status := 'HEALTHY: ' || v_total_sessions || ' total, ' || v_active_sessions || ' active';
   END IF;
   
   RETURN v_health_status;
END;
/

-- Test health check
SELECT connection_pool_health_check() as pool_status FROM dual;

-- Advanced session monitoring query
SELECT s.sid, s.serial#, s.username, s.program, s.machine,
      s.status, s.last_call_et/60 as idle_minutes,
      ss.value as logical_reads,
      ROUND(ss.value / GREATEST(s.last_call_et, 1), 2) as reads_per_second
FROM v$session s
LEFT JOIN v$sesstat ss ON s.sid = ss.sid 
   AND ss.statistic# = (SELECT statistic# FROM v$statname WHERE name = 'session logical reads')
WHERE s.username IS NOT NULL
ORDER BY ss.value DESC;

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

Data state before query executes.

Output / Afterresult

After

DYNAMIC_SQL	EXECUTED	ROWS_AFFECTED
UPDATE employees SET...	✅	2 rows
DROP TABLE temp_tbl	✅	Table removed

EXECUTE IMMEDIATE runs SQL assembled as a string at runtime

Key Takeaway

Use Oracle connection pooling features, DRCP (Database Resident Connection Pooling), and proper session management techniques for optimal performance.

Q168. Write a query to implement real-time data synchronization between Oracle databases.

Answer

Use Oracle Streams, GoldenGate, or custom triggers and database links to implement real-time data replication and synchronization.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

CodeOracle SQL

SQL

-- Method 1: Database Link and Trigger-based Synchronization
-- Create database link to target database
CREATE DATABASE LINK target_db
CONNECT TO sync_user IDENTIFIED BY password
USING 'target_database_tns';

-- Test database link
SELECT * FROM dual@target_db;

-- Create synchronization log table
CREATE TABLE sync_log (
   log_id NUMBER GENERATED ALWAYS AS IDENTITY,
   table_name VARCHAR2(50),
   operation VARCHAR2(10),
   record_id VARCHAR2(100),
   sync_status VARCHAR2(20) DEFAULT 'PENDING',
   error_message VARCHAR2(4000),
   created_date TIMESTAMP DEFAULT SYSTIMESTAMP,
   processed_date TIMESTAMP
);

-- Synchronization trigger for employees table
CREATE OR REPLACE TRIGGER trg_employee_sync
AFTER INSERT OR UPDATE OR DELETE ON employees
FOR EACH ROW
DECLARE
   v_operation VARCHAR2(10);
   v_employee_id NUMBER;
   v_sql VARCHAR2(4000);
BEGIN
   -- Determine operation and employee ID
   IF INSERTING THEN
       v_operation := 'INSERT';
       v_employee_id := :NEW.employee_id;
       
       v_sql := 'INSERT INTO employees@target_db (employee_id, first_name, last_name, email, salary, department_id, hire_date) ' ||
                'VALUES (' || :NEW.employee_id || ', ''' || :NEW.first_name || ''', ''' || :NEW.last_name || ''', ''' ||
                :NEW.email || ''', ' || :NEW.salary || ', ' || :NEW.department_id || ', DATE ''' || 
                TO_CHAR(:NEW.hire_date, 'YYYY-MM-DD') || ''')';
                
   ELSIF UPDATING THEN
       v_operation := 'UPDATE';
       v_employee_id := :NEW.employee_id;
       
       v_sql := 'UPDATE employees@target_db SET first_name = ''' || :NEW.first_name || 
                ''', last_name = ''' || :NEW.last_name || ''', email = ''' || :NEW.email ||
                ''', salary = ' || :NEW.salary || ', department_id = ' || :NEW.department_id ||
                ', hire_date = DATE ''' || TO_CHAR(:NEW.hire_date, 'YYYY-MM-DD') || 
                ''' WHERE employee_id = ' || :NEW.employee_id;
                
   ELSIF DELETING THEN
       v_operation := 'DELETE';
       v_employee_id := :OLD.employee_id;
       
       v_sql := 'DELETE FROM employees@target_db WHERE employee_id = ' || :OLD.employee_id;
   END IF;
   
   -- Log synchronization request
   INSERT INTO sync_log (table_name, operation, record_id, sync_status)
   VALUES ('EMPLOYEES', v_operation, v_employee_id, 'PENDING');
   
   -- Execute synchronization
   BEGIN
       EXECUTE IMMEDIATE v_sql;
       
       -- Update log as successful
       UPDATE sync_log 
       SET sync_status = 'SUCCESS', processed_date = SYSTIMESTAMP
       WHERE log_id = (SELECT MAX(log_id) FROM sync_log WHERE record_id = v_employee_id);
       
   EXCEPTION
       WHEN OTHERS THEN
           -- Log error
           UPDATE sync_log 
           SET sync_status = 'ERROR', 
               error_message = SQLERRM,
               processed_date = SYSTIMESTAMP
           WHERE log_id = (SELECT MAX(log_id) FROM sync_log WHERE record_id = v_employee_id);
   END;
END;
/

-- Method 2: Batch Synchronization Procedure
CREATE OR REPLACE PROCEDURE sync_employees_batch IS
   CURSOR unsync_records IS
       SELECT log_id, table_name, operation, record_id
       FROM sync_log
       WHERE sync_status = 'PENDING'
         AND table_name = 'EMPLOYEES'
       ORDER BY created_date;
   
   v_sql VARCHAR2(4000);
   v_employee_rec employees%ROWTYPE;
BEGIN
   FOR rec IN unsync_records LOOP
       BEGIN
           -- Get current employee data
           SELECT * INTO v_employee_rec
           FROM employees
           WHERE employee_id = rec.record_id;
           
           -- Build sync SQL based on operation
           CASE rec.operation
               WHEN 'INSERT' THEN
                   v_sql := 'INSERT INTO employees@target_db VALUES (' ||
                            v_employee_rec.employee_id || ', ''' ||
                            v_employee_rec.first_name || ''', ''' ||
                            v_employee_rec.last_name || ''', ''' ||
                            v_employee_rec.email || ''', ' ||
                            v_employee_rec.salary || ', ' ||
                            v_employee_rec.department_id || ', DATE ''' ||
                            TO_CHAR(v_employee_rec.hire_date, 'YYYY-MM-DD') || ''')';
                            
               WHEN 'UPDATE' THEN
                   v_sql := 'UPDATE employees@target_db SET ' ||
                            'first_name = ''' || v_employee_rec.first_name || ''', ' ||
                            'last_name = ''' || v_employee_rec.last_name || ''', ' ||
                            'email = ''' || v_employee_rec.email || ''', ' ||
                            'salary = ' || v_employee_rec.salary || ', ' ||
                            'department_id = ' || v_employee_rec.department_id || ', ' ||
                            'hire_date = DATE ''' || TO_CHAR(v_employee_rec.hire_date, 'YYYY-MM-DD') ||
                            ''' WHERE employee_id = ' || v_employee_rec.employee_id;
                            
               WHEN 'DELETE' THEN
                   v_sql := 'DELETE FROM employees@target_db WHERE employee_id = ' || rec.record_id;
           END CASE;
           
           -- Execute synchronization
           EXECUTE IMMEDIATE v_sql;
           
           -- Update sync status
           UPDATE sync_log
           SET sync_status = 'SUCCESS', processed_date = SYSTIMESTAMP
           WHERE log_id = rec.log_id;
           
       EXCEPTION
           WHEN NO_DATA_FOUND THEN
               -- Handle deleted records
               IF rec.operation != 'DELETE' THEN
                   UPDATE sync_log
                   SET sync_status = 'ERROR',
                       error_message = 'Record not found in source table',
                       processed_date = SYSTIMESTAMP
                   WHERE log_id = rec.log_id;
               END IF;
               
           WHEN OTHERS THEN
               UPDATE sync_log
               SET sync_status = 'ERROR',
                   error_message = SQLERRM,
                   processed_date = SYSTIMESTAMP
               WHERE log_id = rec.log_id;
       END;
       
       COMMIT;
   END LOOP;
END;
/

-- Method 3: Conflict Resolution for Bi-directional Sync
CREATE TABLE sync_conflicts (
   conflict_id NUMBER GENERATED ALWAYS AS IDENTITY,
   table_name VARCHAR2(50),
   record_id VARCHAR2(100),
   source_database VARCHAR2(50),
   target_database VARCHAR2(50),
   conflict_type VARCHAR2(50),
   source_data CLOB,
   target_data CLOB,
   resolution_status VARCHAR2(20) DEFAULT 'UNRESOLVED',
   created_date TIMESTAMP DEFAULT SYSTIMESTAMP,
   resolved_date TIMESTAMP
);

-- Conflict detection procedure
CREATE OR REPLACE PROCEDURE detect_sync_conflicts IS
   CURSOR source_records IS
       SELECT employee_id, first_name, last_name, email, salary, 
              department_id, hire_date
       FROM employees;
   
   v_target_rec employees%ROWTYPE;
   v_conflict_detected BOOLEAN;
BEGIN
   FOR src_rec IN source_records LOOP
       v_conflict_detected := FALSE;
       
       BEGIN
           -- Get corresponding target record
           SELECT * INTO v_target_rec
           FROM employees@target_db
           WHERE employee_id = src_rec.employee_id;
           
           -- Check for data conflicts
           IF src_rec.first_name != v_target_rec.first_name OR
              src_rec.last_name != v_target_rec.last_name OR
              src_rec.email != v_target_rec.email OR
              src_rec.salary != v_target_rec.salary OR
              src_rec.department_id != v_target_rec.department_id THEN
               
               v_conflict_detected := TRUE;
               
               -- Log conflict
               INSERT INTO sync_conflicts (
                   table_name, record_id, source_database, target_database,
                   conflict_type, source_data, target_data
               ) VALUES (
                   'EMPLOYEES', src_rec.employee_id, 'SOURCE_DB', 'TARGET_DB',
                   'DATA_MISMATCH',
                   JSON_OBJECT(
                       'first_name' VALUE src_rec.first_name,
                       'last_name' VALUE src_rec.last_name,
                       'email' VALUE src_rec.email,
                       'salary' VALUE src_rec.salary,
                       'department_id' VALUE src_rec.department_id
                   ),
                   JSON_OBJECT(
                       'first_name' VALUE v_target_rec.first_name,
                       'last_name' VALUE v_target_rec.last_name,
                       'email' VALUE v_target_rec.email,
                       'salary' VALUE v_target_rec.salary,
                       'department_id' VALUE v_target_rec.department_id
                   )
               );
           END IF;
           
       EXCEPTION
           WHEN NO_DATA_FOUND THEN
               -- Record exists in source but not in target
               INSERT INTO sync_conflicts (
                   table_name, record_id, source_database, target_database,
                   conflict_type, source_data
               ) VALUES (
                   'EMPLOYEES', src_rec.employee_id, 'SOURCE_DB', 'TARGET_DB',
                   'MISSING_IN_TARGET',
                   JSON_OBJECT(
                       'first_name' VALUE src_rec.first_name,
                       'last_name' VALUE src_rec.last_name,
                       'email' VALUE src_rec.email,
                       'salary' VALUE src_rec.salary,
                       'department_id' VALUE src_rec.department_id
                   )
               );
       END;
   END LOOP;
   
   COMMIT;
END;
/

-- Method 4: Monitoring and Alerting
CREATE OR REPLACE PROCEDURE monitor_sync_health IS
   v_pending_count NUMBER;
   v_error_count NUMBER;
   v_old_errors NUMBER;
   v_alert_message VARCHAR2(4000);
BEGIN
   -- Count pending synchronizations
   SELECT COUNT(*) INTO v_pending_count
   FROM sync_log
   WHERE sync_status = 'PENDING'
     AND created_date >= SYSTIMESTAMP - INTERVAL '1' HOUR;
   
   -- Count recent errors
   SELECT COUNT(*) INTO v_error_count
   FROM sync_log
   WHERE sync_status = 'ERROR'
     AND created_date >= SYSTIMESTAMP - INTERVAL '1' HOUR;
   
   -- Count old unresolved errors
   SELECT COUNT(*) INTO v_old_errors
   FROM sync_log
   WHERE sync_status = 'ERROR'
     AND created_date < SYSTIMESTAMP - INTERVAL '24' HOUR;
   
   -- Generate alerts
   IF v_pending_count > 100 THEN
       v_alert_message := 'HIGH: ' || v_pending_count || ' pending sync operations';
       -- Send alert (implement your alerting mechanism)
       DBMS_OUTPUT.PUT_LINE(v_alert_message);
   END IF;
   
   IF v_error_count > 10 THEN
       v_alert_message := 'CRITICAL: ' || v_error_count || ' sync errors in last hour';
       DBMS_OUTPUT.PUT_LINE(v_alert_message);
   END IF;
   
   IF v_old_errors > 0 THEN
       v_alert_message := 'WARNING: ' || v_old_errors || ' unresolved sync errors older than 24 hours';
       DBMS_OUTPUT.PUT_LINE(v_alert_message);
   END IF;
   
   -- Log health check
   INSERT INTO sync_health_log (
       check_date, pending_count, error_count, old_error_count
   ) VALUES (
       SYSTIMESTAMP, v_pending_count, v_error_count, v_old_errors
   );
   
   COMMIT;
END;
/

-- Create health monitoring table
CREATE TABLE sync_health_log (
   check_date TIMESTAMP,
   pending_count NUMBER,
   error_count NUMBER,
   old_error_count NUMBER
);

-- Schedule health monitoring job
BEGIN
   DBMS_SCHEDULER.CREATE_JOB(
       job_name => 'SYNC_HEALTH_MONITOR',
       job_type => 'PLSQL_BLOCK',
       job_action => 'BEGIN monitor_sync_health; END;',
       start_date => SYSTIMESTAMP,
       repeat_interval => 'FREQ=MINUTELY; INTERVAL=15',
       enabled => TRUE,
       comments => 'Monitor synchronization health every 15 minutes'
   );
END;
/

-- Sync status queries
-- Overall sync performance
SELECT sync_status, COUNT(*) as count,
      MIN(created_date) as oldest_request,
      MAX(created_date) as newest_request
FROM sync_log
WHERE created_date >= TRUNC(SYSDATE)
GROUP BY sync_status;

-- Error analysis
SELECT error_message, COUNT(*) as frequency
FROM sync_log
WHERE sync_status = 'ERROR'
 AND created_date >= TRUNC(SYSDATE) - 7
GROUP BY error_message
ORDER BY COUNT(*) DESC;

-- Sync latency analysis
SELECT table_name,
      AVG(EXTRACT(SECOND FROM (processed_date - created_date))) as avg_latency_seconds,
      MAX(EXTRACT(SECOND FROM (processed_date - created_date))) as max_latency_seconds,
      COUNT(*) as total_operations
FROM sync_log
WHERE sync_status = 'SUCCESS'
 AND created_date >= TRUNC(SYSDATE) - 1
GROUP BY table_name;

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

Data state before query executes.

Output / Afterresult

After

DYNAMIC_SQL	EXECUTED	ROWS_AFFECTED
UPDATE employees SET...	✅	2 rows
DROP TABLE temp_tbl	✅	Table removed

EXECUTE IMMEDIATE runs SQL assembled as a string at runtime

Key Takeaway

Use Oracle Streams, GoldenGate, or custom triggers and database links to implement real-time data replication and synchronization.

Oracle Data Modeling & Star Schema

Q169. How do you implement a slowly changing dimension (SCD) Type 2 in Oracle?

Answer

SCD Type 2 maintains historical data by creating new records for changes, using effective dates and flags to track current vs historical records.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

CodeOracle SQL

SQL

-- Create SCD Type 2 dimension table
CREATE TABLE dim_customer (
   customer_key NUMBER GENERATED ALWAYS AS IDENTITY,
   customer_id NUMBER NOT NULL,
   customer_name VARCHAR2(100),
   email VARCHAR2(100),
   phone VARCHAR2(20),
   address VARCHAR2(200),
   effective_start_date DATE DEFAULT SYSDATE,
   effective_end_date DATE DEFAULT DATE '9999-12-31',
   is_current_flag CHAR(1) DEFAULT 'Y',
   created_date DATE DEFAULT SYSDATE,
   PRIMARY KEY (customer_key)
);

-- Create indexes for performance
CREATE INDEX idx_dim_customer_id ON dim_customer(customer_id, is_current_flag);
CREATE INDEX idx_dim_customer_dates ON dim_customer(effective_start_date, effective_end_date);

-- SCD Type 2 Merge procedure
CREATE OR REPLACE PROCEDURE update_customer_scd(
   p_customer_id NUMBER,
   p_customer_name VARCHAR2,
   p_email VARCHAR2,
   p_phone VARCHAR2,
   p_address VARCHAR2
) IS
   v_current_record dim_customer%ROWTYPE;
   v_record_changed BOOLEAN := FALSE;
BEGIN
   -- Get current record
   SELECT * INTO v_current_record
   FROM dim_customer
   WHERE customer_id = p_customer_id
     AND is_current_flag = 'Y';
   
   -- Check if any tracked attributes changed
   IF NVL(v_current_record.customer_name, 'NULL') != NVL(p_customer_name, 'NULL') OR
      NVL(v_current_record.email, 'NULL') != NVL(p_email, 'NULL') OR
      NVL(v_current_record.phone, 'NULL') != NVL(p_phone, 'NULL') OR
      NVL(v_current_record.address, 'NULL') != NVL(p_address, 'NULL') THEN
       v_record_changed := TRUE;
   END IF;
   
   IF v_record_changed THEN
       -- Expire current record
       UPDATE dim_customer
       SET effective_end_date = SYSDATE - INTERVAL '1' SECOND,
           is_current_flag = 'N'
       WHERE customer_key = v_current_record.customer_key;
       
       -- Insert new current record
       INSERT INTO dim_customer (
           customer_id, customer_name, email, phone, address,
           effective_start_date, effective_end_date, is_current_flag
       ) VALUES (
           p_customer_id, p_customer_name, p_email, p_phone, p_address,
           SYSDATE, DATE '9999-12-31', 'Y'
       );
   END IF;
   
   COMMIT;
EXCEPTION
   WHEN NO_DATA_FOUND THEN
       -- Insert new customer
       INSERT INTO dim_customer (
           customer_id, customer_name, email, phone, address
       ) VALUES (
           p_customer_id, p_customer_name, p_email, p_phone, p_address
       );
       COMMIT;
END;
/

-- Query current records only
SELECT customer_id, customer_name, email, phone
FROM dim_customer
WHERE is_current_flag = 'Y';

-- Query historical changes for a customer
SELECT customer_id, customer_name, email,
      effective_start_date, effective_end_date, is_current_flag
FROM dim_customer
WHERE customer_id = 12345
ORDER BY effective_start_date;

-- Point-in-time query (what did customer look like on specific date)
SELECT customer_id, customer_name, email, phone, address
FROM dim_customer
WHERE customer_id = 12345
 AND DATE '2023-06-15' BETWEEN effective_start_date AND effective_end_date;

-- Bulk SCD processing using MERGE
MERGE INTO dim_customer d
USING (
   SELECT customer_id, customer_name, email, phone, address
   FROM staging_customer_updates
) s ON (d.customer_id = s.customer_id AND d.is_current_flag = 'Y')
WHEN MATCHED THEN
   UPDATE SET 
       effective_end_date = CASE 
           WHEN (NVL(d.customer_name,'X') != NVL(s.customer_name,'X') OR
                 NVL(d.email,'X') != NVL(s.email,'X') OR
                 NVL(d.phone,'X') != NVL(s.phone,'X') OR
                 NVL(d.address,'X') != NVL(s.address,'X'))
           THEN SYSDATE - INTERVAL '1' SECOND
           ELSE d.effective_end_date
       END,
       is_current_flag = CASE 
           WHEN (NVL(d.customer_name,'X') != NVL(s.customer_name,'X') OR
                 NVL(d.email,'X') != NVL(s.email,'X') OR
                 NVL(d.phone,'X') != NVL(s.phone,'X') OR
                 NVL(d.address,'X') != NVL(s.address,'X'))
           THEN 'N'
           ELSE 'Y'
       END;

-- Insert new versions for changed records
INSERT INTO dim_customer (customer_id, customer_name, email, phone, address)
SELECT s.customer_id, s.customer_name, s.email, s.phone, s.address
FROM staging_customer_updates s
JOIN dim_customer d ON s.customer_id = d.customer_id
WHERE d.is_current_flag = 'N'
 AND d.effective_end_date = SYSDATE - INTERVAL '1' SECOND;

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

Data state before query executes.

Output / Afterresult

After

EMP_ID	EMP_NAME	ACTION	RESULT
101	Alice Johnson	UPDATED (matched)	Salary → 55,000
102	Bob Smith	UPDATED (matched)	Salary → 68,200
999	New Hire	INSERTED (not matched)	New row added

MERGE: 2 rows updated + 1 inserted in one atomic statement

Key Takeaway

SCD Type 2 maintains historical data by creating new records for changes, using effective dates and flags to track current vs historical records.

Q170. Write a query to implement a data warehouse star schema with fact and dimension tables.

Answer

Design and query a star schema with fact tables containing measures and foreign keys to dimension tables containing descriptive attributes.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

CodeOracle SQL

SQL

-- Create dimension tables for star schema

-- Time Dimension
CREATE TABLE dim_time (
   time_key NUMBER PRIMARY KEY,
   date_value DATE NOT NULL,
   year NUMBER,
   quarter NUMBER,
   month NUMBER,
   month_name VARCHAR2(20),
   day_of_month NUMBER,
   day_of_week NUMBER,
   day_name VARCHAR2(20),
   week_of_year NUMBER,
   is_weekend CHAR(1),
   is_holiday CHAR(1),
   fiscal_year NUMBER,
   fiscal_quarter NUMBER
);

-- Product Dimension  
CREATE TABLE dim_product (
   product_key NUMBER PRIMARY KEY,
   product_id VARCHAR2(20) UNIQUE,
   product_name VARCHAR2(100),
   brand VARCHAR2(50),
   category VARCHAR2(50),
   subcategory VARCHAR2(50),
   unit_cost NUMBER(10,2),
   unit_price NUMBER(10,2),
   product_status VARCHAR2(20),
   launch_date DATE
);

-- Customer Dimension
CREATE TABLE dim_customer (
   customer_key NUMBER PRIMARY KEY,
   customer_id VARCHAR2(20) UNIQUE,
   customer_name VARCHAR2(100),
   customer_type VARCHAR2(20),
   segment VARCHAR2(30),
   region VARCHAR2(50),
   country VARCHAR2(50),
   city VARCHAR2(50),
   zip_code VARCHAR2(10),
   registration_date DATE,
   credit_rating VARCHAR2(10)
);

-- Store Dimension
CREATE TABLE dim_store (
   store_key NUMBER PRIMARY KEY,
   store_id VARCHAR2(20) UNIQUE,
   store_name VARCHAR2(100),
   store_type VARCHAR2(30),
   store_size VARCHAR2(20),
   region VARCHAR2(50),
   district VARCHAR2(50),
   city VARCHAR2(50),
   state VARCHAR2(50),
   country VARCHAR2(50),
   open_date DATE,
   manager_name VARCHAR2(100)
);

-- Fact Table - Sales Facts
CREATE TABLE fact_sales (
   sales_key NUMBER PRIMARY KEY,
   time_key NUMBER REFERENCES dim_time(time_key),
   product_key NUMBER REFERENCES dim_product(product_key),
   customer_key NUMBER REFERENCES dim_customer(customer_key),
   store_key NUMBER REFERENCES dim_store(store_key),
   -- Measures
   quantity_sold NUMBER,
   unit_price NUMBER(10,2),
   unit_cost NUMBER(10,2),
   sales_amount NUMBER(12,2),
   discount_amount NUMBER(10,2),
   profit_amount NUMBER(12,2),
   tax_amount NUMBER(10,2)
);

-- Create indexes for performance
CREATE INDEX idx_fact_sales_time ON fact_sales(time_key);
CREATE INDEX idx_fact_sales_product ON fact_sales(product_key);
CREATE INDEX idx_fact_sales_customer ON fact_sales(customer_key);
CREATE INDEX idx_fact_sales_store ON fact_sales(store_key);

-- Composite index for common query patterns
CREATE INDEX idx_fact_sales_time_store ON fact_sales(time_key, store_key);
CREATE INDEX idx_fact_sales_time_product ON fact_sales(time_key, product_key);

-- Populate dimension tables with sample data
-- Time dimension population
INSERT INTO dim_time (time_key, date_value, year, quarter, month, month_name, 
                    day_of_month, day_of_week, day_name, week_of_year, 
                    is_weekend, is_holiday, fiscal_year, fiscal_quarter)
SELECT ROWNUM as time_key,
      date_value,
      EXTRACT(YEAR FROM date_value) as year,
      CEIL(EXTRACT(MONTH FROM date_value) / 3) as quarter,
      EXTRACT(MONTH FROM date_value) as month,
      TO_CHAR(date_value, 'Month') as month_name,
      EXTRACT(DAY FROM date_value) as day_of_month,
      TO_NUMBER(TO_CHAR(date_value, 'D')) as day_of_week,
      TO_CHAR(date_value, 'Day') as day_name,
      TO_NUMBER(TO_CHAR(date_value, 'WW')) as week_of_year,
      CASE WHEN TO_CHAR(date_value, 'D') IN ('1', '7') THEN 'Y' ELSE 'N' END as is_weekend,
      'N' as is_holiday, -- Simplified
      CASE WHEN EXTRACT(MONTH FROM date_value) >= 4 
           THEN EXTRACT(YEAR FROM date_value) 
           ELSE EXTRACT(YEAR FROM date_value) - 1 END as fiscal_year,
      CASE WHEN EXTRACT(MONTH FROM date_value) IN (4,5,6) THEN 1
           WHEN EXTRACT(MONTH FROM date_value) IN (7,8,9) THEN 2
           WHEN EXTRACT(MONTH FROM date_value) IN (10,11,12) THEN 3
           ELSE 4 END as fiscal_quarter
FROM (
   SELECT DATE '2022-01-01' + LEVEL - 1 as date_value
   FROM dual
   CONNECT BY LEVEL <= 1095 -- 3 years of data
);

-- Star Schema Analytical Queries

-- 1. Sales by Year and Quarter
SELECT dt.year, dt.quarter,
      SUM(fs.sales_amount) as total_sales,
      SUM(fs.profit_amount) as total_profit,
      COUNT(*) as transaction_count,
      ROUND(AVG(fs.sales_amount), 2) as avg_transaction_value
FROM fact_sales fs
JOIN dim_time dt ON fs.time_key = dt.time_key
GROUP BY dt.year, dt.quarter
ORDER BY dt.year, dt.quarter;

-- 2. Top Products by Sales in Each Category
WITH product_sales AS (
   SELECT dp.category, dp.product_name, dp.brand,
          SUM(fs.sales_amount) as total_sales,
          SUM(fs.quantity_sold) as total_quantity,
          ROW_NUMBER() OVER (PARTITION BY dp.category ORDER BY SUM(fs.sales_amount) DESC) as rn
   FROM fact_sales fs
   JOIN dim_product dp ON fs.product_key = dp.product_key
   JOIN dim_time dt ON fs.time_key = dt.time_key
   WHERE dt.year = 2023
   GROUP BY dp.category, dp.product_name, dp.brand
)
SELECT category, product_name, brand, total_sales, total_quantity
FROM product_sales
WHERE rn <= 3
ORDER BY category, rn;

-- 3. Customer Segmentation Analysis
SELECT dc.segment, dc.region,
      COUNT(DISTINCT dc.customer_key) as customer_count,
      SUM(fs.sales_amount) as total_sales,
      ROUND(AVG(fs.sales_amount), 2) as avg_order_value,
      SUM(fs.quantity_sold) as total_units,
      ROUND(SUM(fs.sales_amount) / COUNT(DISTINCT dc.customer_key), 2) as sales_per_customer
FROM fact_sales fs
JOIN dim_customer dc ON fs.customer_key = dc.customer_key
JOIN dim_time dt ON fs.time_key = dt.time_key
WHERE dt.year = 2023
GROUP BY dc.segment, dc.region
ORDER BY total_sales DESC;

-- 4. Store Performance Analysis
SELECT ds.region, ds.store_name, ds.store_type,
      SUM(fs.sales_amount) as total_sales,
      SUM(fs.profit_amount) as total_profit,
      ROUND(SUM(fs.profit_amount) / SUM(fs.sales_amount) * 100, 2) as profit_margin_pct,
      COUNT(DISTINCT fs.customer_key) as unique_customers,
      COUNT(*) as transaction_count
FROM fact_sales fs
JOIN dim_store ds ON fs.store_key = ds.store_key
JOIN dim_time dt ON fs.time_key = dt.time_key
WHERE dt.year = 2023
GROUP BY ds.region, ds.store_name, ds.store_type
ORDER BY total_sales DESC;

-- 5. Time Series Analysis - Monthly Trends
SELECT dt.year, dt.month, dt.month_name,
      SUM(fs.sales_amount) as monthly_sales,
      LAG(SUM(fs.sales_amount)) OVER (ORDER BY dt.year, dt.month) as prev_month_sales,
      ROUND(
          (SUM(fs.sales_amount) - LAG(SUM(fs.sales_amount)) OVER (ORDER BY dt.year, dt.month)) /
          LAG(SUM(fs.sales_amount)) OVER (ORDER BY dt.year, dt.month) * 100, 2
      ) as mom_growth_pct,
      SUM(SUM(fs.sales_amount)) OVER (PARTITION BY dt.year ORDER BY dt.month) as ytd_sales
FROM fact_sales fs
JOIN dim_time dt ON fs.time_key = dt.time_key
GROUP BY dt.year, dt.month, dt.month_name
ORDER BY dt.year, dt.month;

-- 6. Complex Drill-Down Query
SELECT dt.year, dt.quarter, dc.region, dp.category,
      SUM(fs.sales_amount) as sales,
      SUM(fs.profit_amount) as profit,
      COUNT(*) as transactions,
      -- Ranking within each time period
      RANK() OVER (PARTITION BY dt.year, dt.quarter ORDER BY SUM(fs.sales_amount) DESC) as sales_rank,
      -- Percentage of total sales
      ROUND(SUM(fs.sales_amount) / SUM(SUM(fs.sales_amount)) OVER (PARTITION BY dt.year, dt.quarter) * 100, 2) as pct_of_period_sales
FROM fact_sales fs
JOIN dim_time dt ON fs.time_key = dt.time_key
JOIN dim_customer dc ON fs.customer_key = dc.customer_key
JOIN dim_product dp ON fs.product_key = dp.product_key
WHERE dt.year BETWEEN 2022 AND 2023
GROUP BY dt.year, dt.quarter, dc.region, dp.category
HAVING SUM(fs.sales_amount) > 10000
ORDER BY dt.year, dt.quarter, sales DESC;

-- 7. Cube and Rollup Operations
SELECT dt.year, dp.category, dc.segment,
      SUM(fs.sales_amount) as total_sales,
      COUNT(*) as transaction_count,
      GROUPING_ID(dt.year, dp.category, dc.segment) as grouping_level
FROM fact_sales fs
JOIN dim_time dt ON fs.time_key = dt.time_key
JOIN dim_product dp ON fs.product_key = dp.product_key
JOIN dim_customer dc ON fs.customer_key = dc.customer_key
WHERE dt.year = 2023
GROUP BY CUBE(dt.year, dp.category, dc.segment)
ORDER BY grouping_level, dt.year, dp.category, dc.segment;

-- 8. Performance Optimization with Materialized Views
CREATE MATERIALIZED VIEW mv_monthly_sales
BUILD IMMEDIATE
REFRESH FAST ON COMMIT
AS
SELECT dt.year, dt.month, dp.category, dc.region,
      SUM(fs.sales_amount) as total_sales,
      SUM(fs.profit_amount) as total_profit,
      COUNT(*) as transaction_count,
      COUNT(DISTINCT fs.customer_key) as unique_customers
FROM fact_sales fs
JOIN dim_time dt ON fs.time_key = dt.time_key
JOIN dim_product dp ON fs.product_key = dp.product_key
JOIN dim_customer dc ON fs.customer_key = dc.customer_key
GROUP BY dt.year, dt.month, dp.category, dc.region;

-- Create index on materialized view
CREATE INDEX idx_mv_monthly_sales ON mv_monthly_sales(year, month, category, region);

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

Data state before query executes.

Output / Afterresult

After

EMP_NAME	DEPT_NAME
Alice Johnson	Engineering
Bob Smith	Marketing
Carol White	Engineering

3 rows — only rows with matching dept_id in both tables (Dave Brown excluded — no dept)

Key Takeaway

Design and query a star schema with fact tables containing measures and foreign keys to dimension tables containing descriptive attributes.

Miscellaneous Advanced Topics

Q171. Discuss the use of dynamic SQL.

Q172. How do you implement full-text search in SQL databases?

Answer

Implementing full-text search in SQL databases involves utilizing SQL functions like "CONTAINS" or "FREETEXT," creating full-text indexes, and considering the integration of third-party search engines for more advanced and efficient text searching capabilities.

These techniques enable SQL developers to enhance the search functionality of their database applications, making it easier to retrieve relevant information from large volumes of text data.One fundamental approach is to use the "CONTAINS" or "FREETEXT" functions, depending on the database system being used, such as SQL Server or PostgreSQL.

These functions enable users to perform text searches on designated columns, returning results that match the specified keywords or phrases.

Database systems optimize these search operations for improved performance, by creating full-text indexes on the columns of interest.SQL offers the "LIKE" operator, which allows for pattern matching within text data. "LIKE" is used for basic text searching by specifying wildcard characters to match partial text strings, which allows for pattern matching within text data.SQL databases support the integration of third-party search engines like Elasticsearch or Solr, which are specifically designed for advanced full-text search capabilities.

These engines offer powerful features such as relevance ranking, stemming, and faceted search, enhancing the accuracy and flexibility of text searches within SQL databases.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

CodeOracle SQL

SQL

SELECT * FROM employees WHERE ROWNUM <= 5;

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

Data state before query executes.

Output / Afterresult

After

EMP_ID	EMP_NAME	SALARY
101	Alice Johnson	50,000
102	Bob Smith	62,000
103	Carol White	55,000

Top 3 rows — ROWNUM/FETCH FIRST limits result set

Key Takeaway

Q173. Explain temporal tables and their uses.

Q174. Explain the concept and implementation of partitioned tables in SQL.

Q175. Discuss the use and challenges of managing temporal data in SQL.

Q176. How do you implement and manage table inheritance in SQL?

Q177. What are the challenges and solutions for managing multilingual data in SQL?

Answer

The challenges in managing multilingual data in SQL involve handling various character sets, ensuring proper collation for sorting and comparing data, and dealing with issues related to case sensitivity and accent marks.

SQL databases use specific character sets like UTF-8 to support a wide range of languages.

SQL allows the setting of collation at the database, table, or column level for accurate sorting and comparison of multilingual data.

This ensures that data is sorted and compared according to the linguistic rules of the relevant language.To handle case sensitivity and accent differences in multilingual data, SQL provides collation settings that can be fine-tuned for case-insensitive and accent-insensitive comparisons.

Indexes should be created on columns containing multilingual data to improve query performance.

Proper normalization of the database schema is crucial to avoid redundancy and ensure efficient storage of multilingual data.

Utilize appropriate data types like NVARCHAR for storing multilingual content to handle the extended character sets effectively.

Regular database maintenance and updates ensure the continued compatibility and performance of multilingual data handling in SQL environments.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

CodeOracle SQL

SQL

SELECT * FROM employees WHERE ROWNUM <= 5;

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

Data state before query executes.

Output / Afterresult

After

EMP_ID	EMP_NAME	SALARY
101	Alice Johnson	50,000
102	Bob Smith	62,000
103	Carol White	55,000

Top 3 rows — ROWNUM/FETCH FIRST limits result set

Key Takeaway

Q178. Explain the concept of virtual columns in SQL and their uses.

Q179. How do you manage version control for database schema changes?

Answer

To manage version control for database schema changes, specialized tools like .

These tools, like Liquibase or Flyway are used.

These tools track and manage database schema changes using a version control system.

This approach ensures that schema changes are consistent across different environments and team members.

The version control system acts like a repository for SQL scripts, which are written to modify the database schema.

Developers write SQL migration scripts for each change, and the version control tool applies these scripts in the correct order.The process involves maintaining an up-to-date schema version in the database, in a dedicated table.

This table logs the history of applied migrations, allowing for rollback of changes if necessary.

The version control tool compares the database's current schema version with the available migration scripts, when deploying new changes.

It then executes the pending scripts to update the schema.

This method guarantees that schema changes are applied in a controlled and predictable manner, reducing the risk of conflicts or inconsistencies.

Integrating database version control with an overall code version control system aligns database changes with corresponding application code changes, ensuring that they are in sync.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

CodeOracle SQL

SQL

SELECT * FROM employees WHERE ROWNUM <= 5;

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

Data state before query executes.

Output / Afterresult

After

EMP_ID	EMP_NAME	SALARY
101	Alice Johnson	50,000
102	Bob Smith	62,000
103	Carol White	55,000

Top 3 rows — ROWNUM/FETCH FIRST limits result set

Key Takeaway

To manage version control for database schema changes, specialized tools like .

Q180. How do you implement temporal tables and track data changes over time in Oracle?

Answer

Use Oracle Temporal Validity and Flashback Data Archive features to track historical changes and implement temporal queries for time-based data analysis.

Sample Dataused in example

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

CodeOracle SQL

SQL

-- Method 1: Using Temporal Validity (Oracle 12c+)
CREATE TABLE employee_temporal (
   employee_id NUMBER,
   first_name VARCHAR2(50),
   last_name VARCHAR2(50),
   salary NUMBER,
   department_id NUMBER,
   -- Temporal validity columns
   valid_from DATE,
   valid_to DATE,
   PERIOD FOR valid_time (valid_from, valid_to)
);

-- Enable temporal validity
ALTER TABLE employee_temporal ADD (
   CONSTRAINT emp_temporal_pk PRIMARY KEY (employee_id, valid_time WITHOUT OVERLAPS)
);

-- Insert temporal data
INSERT INTO employee_temporal (employee_id, first_name, last_name, salary, department_id, valid_from, valid_to)
VALUES (101, 'John', 'Doe', 50000, 10, DATE '2023-01-01', DATE '2023-06-30');

INSERT INTO employee_temporal (employee_id, first_name, last_name, salary, department_id, valid_from, valid_to)
VALUES (101, 'John', 'Doe', 55000, 10, DATE '2023-07-01', DATE '2023-12-31');

INSERT INTO employee_temporal (employee_id, first_name, last_name, salary, department_id, valid_from, valid_to)
VALUES (101, 'John', 'Doe', 60000, 20, DATE '2024-01-01', DATE '9999-12-31');

-- Query temporal data for specific time point
SELECT employee_id, first_name, last_name, salary, department_id
FROM employee_temporal
WHERE employee_id = 101
 AND DATE '2023-08-15' BETWEEN valid_from AND valid_to;

-- Query all changes for an employee
SELECT employee_id, salary, department_id, valid_from, valid_to,
      valid_to - valid_from + 1 as period_days
FROM employee_temporal
WHERE employee_id = 101
ORDER BY valid_from;

-- Method 2: Manual Temporal Implementation with Triggers
CREATE TABLE employee_history (
   history_id NUMBER GENERATED ALWAYS AS IDENTITY,
   employee_id NUMBER,
   first_name VARCHAR2(50),
   last_name VARCHAR2(50),
   salary NUMBER,
   department_id NUMBER,
   effective_date DATE DEFAULT SYSDATE,
   end_date DATE DEFAULT DATE '9999-12-31',
   operation_type VARCHAR2(10),
   changed_by VARCHAR2(100) DEFAULT USER,
   changed_timestamp TIMESTAMP DEFAULT SYSTIMESTAMP
);

-- Current employee table
CREATE TABLE employees_current (
   employee_id NUMBER PRIMARY KEY,
   first_name VARCHAR2(50),
   last_name VARCHAR2(50),
   salary NUMBER,
   department_id NUMBER,
   last_updated TIMESTAMP DEFAULT SYSTIMESTAMP
);

-- Temporal trigger for history tracking
CREATE OR REPLACE TRIGGER trg_employee_temporal
AFTER INSERT OR UPDATE OR DELETE ON employees_current
FOR EACH ROW
BEGIN
   IF INSERTING THEN
       INSERT INTO employee_history (
           employee_id, first_name, last_name, salary, department_id, operation_type
       ) VALUES (
           :NEW.employee_id, :NEW.first_name, :NEW.last_name, 
           :NEW.salary, :NEW.department_id, 'INSERT'
       );
   ELSIF UPDATING THEN
       -- End previous record
       UPDATE employee_history
       SET end_date = SYSDATE - INTERVAL '1' SECOND
       WHERE employee_id = :OLD.employee_id
         AND end_date = DATE '9999-12-31';
       
       -- Insert new record
       INSERT INTO employee_history (
           employee_id, first_name, last_name, salary, department_id, operation_type
       ) VALUES (
           :NEW.employee_id, :NEW.first_name, :NEW.last_name, 
           :NEW.salary, :NEW.department_id, 'UPDATE'
       );
   ELSIF DELETING THEN
       -- End current record
       UPDATE employee_history
       SET end_date = SYSDATE,
           operation_type = 'DELETE'
       WHERE employee_id = :OLD.employee_id
         AND end_date = DATE '9999-12-31';
   END IF;
END;
/

-- Method 3: Using Flashback Data Archive (Total Recall)
-- Enable Flashback Data Archive (requires DBA privileges)
/*
CREATE FLASHBACK ARCHIVE fla_employee
TABLESPACE flashback_ts
RETENTION 5 YEAR;

ALTER TABLE employees_current FLASHBACK ARCHIVE fla_employee;
*/

-- Query historical data using Flashback
SELECT employee_id, first_name, last_name, salary, department_id
FROM employees_current
AS OF TIMESTAMP TIMESTAMP '2023-06-15 10:00:00';

-- Query all versions between two timestamps
SELECT employee_id, salary, department_id,
      versions_starttime, versions_endtime, versions_operation
FROM employees_current
VERSIONS BETWEEN TIMESTAMP 
   TIMESTAMP '2023-01-01 00:00:00' AND TIMESTAMP '2023-12-31 23:59:59'
WHERE employee_id = 101
ORDER BY versions_starttime;

-- Method 4: Temporal Analytics Functions
-- Create view for easy temporal queries
CREATE OR REPLACE VIEW v_employee_temporal_analysis AS
SELECT h.*,
      LAG(salary) OVER (PARTITION BY employee_id ORDER BY effective_date) as prev_salary,
      salary - LAG(salary) OVER (PARTITION BY employee_id ORDER BY effective_date) as salary_change,
      LAG(department_id) OVER (PARTITION BY employee_id ORDER BY effective_date) as prev_department,
      end_date - effective_date + 1 as days_in_position,
      ROW_NUMBER() OVER (PARTITION BY employee_id ORDER BY effective_date) as position_sequence
FROM employee_history h;

-- Salary progression analysis
SELECT employee_id, first_name, last_name,
      position_sequence, salary, prev_salary, salary_change,
      ROUND((salary_change / prev_salary) * 100, 2) as salary_increase_pct,
      days_in_position
FROM v_employee_temporal_analysis
WHERE employee_id = 101
 AND prev_salary IS NOT NULL
ORDER BY effective_date;

-- Method 5: Temporal Queries for Business Analysis
-- Employees by salary range over time
WITH temporal_snapshot AS (
   SELECT DATE '2023-01-01' + (LEVEL - 1) * 30 as snapshot_date
   FROM dual
   CONNECT BY LEVEL <= 12 -- Monthly snapshots
),
salary_ranges AS (
   SELECT ts.snapshot_date,
          COUNT(CASE WHEN h.salary < 40000 THEN 1 END) as low_salary,
          COUNT(CASE WHEN h.salary BETWEEN 40000 AND 60000 THEN 1 END) as mid_salary,
          COUNT(CASE WHEN h.salary > 60000 THEN 1 END) as high_salary,
          COUNT(*) as total_employees
   FROM temporal_snapshot ts
   LEFT JOIN employee_history h ON ts.snapshot_date BETWEEN h.effective_date AND h.end_date
   GROUP BY ts.snapshot_date
)
SELECT snapshot_date,
      low_salary, mid_salary, high_salary, total_employees,
      ROUND(low_salary / total_employees * 100, 1) as low_salary_pct,
      ROUND(mid_salary / total_employees * 100, 1) as mid_salary_pct,
      ROUND(high_salary / total_employees * 100, 1) as high_salary_pct
FROM salary_ranges
ORDER BY snapshot_date;

-- Department movement analysis
SELECT h1.employee_id, h1.first_name, h1.last_name,
      h1.department_id as from_dept, h2.department_id as to_dept,
      h1.end_date as move_date,
      h2.effective_date - h1.end_date as gap_days
FROM employee_history h1
JOIN employee_history h2 ON h1.employee_id = h2.employee_id
                        AND h2.effective_date = h1.end_date + 1
WHERE h1.department_id != h2.department_id
ORDER BY h1.employee_id, h1.end_date;

-- Method 6: Temporal Data Validation
CREATE OR REPLACE FUNCTION validate_temporal_data(
   p_employee_id NUMBER
) RETURN VARCHAR2 IS
   v_overlaps NUMBER;
   v_gaps NUMBER;
BEGIN
   -- Check for overlapping periods
   SELECT COUNT(*)
   INTO v_overlaps
   FROM employee_history h1
   JOIN employee_history h2 ON h1.employee_id = h2.employee_id
                            AND h1.history_id != h2.history_id
   WHERE h1.employee_id = p_employee_id
     AND h1.effective_date <= h2.end_date
     AND h1.end_date >= h2.effective_date;
   
   -- Check for gaps in timeline
   SELECT COUNT(*)
   INTO v_gaps
   FROM employee_history h1
   JOIN employee_history h2 ON h1.employee_id = h2.employee_id
                            AND h2.effective_date = (
                                SELECT MIN(effective_date)
                                FROM employee_history
                                WHERE employee_id = h1.employee_id
                                  AND effective_date > h1.end_date
                            )
   WHERE h1.employee_id = p_employee_id
     AND h1.end_date + 1 != h2.effective_date;
   
   IF v_overlaps > 0 THEN
       RETURN 'ERROR: Overlapping periods detected';
   ELSIF v_gaps > 0 THEN
       RETURN 'WARNING: Gaps in timeline detected';
   ELSE
       RETURN 'OK: Timeline is valid';
   END IF;
END;
/

-- Test temporal validation
SELECT validate_temporal_data(101) as validation_result FROM dual;

-- Point-in-time reconstruction
CREATE OR REPLACE FUNCTION get_employee_at_date(p_employee_id NUMBER,
   p_date DATE
) RETURN SYS_REFCURSOR IS
   v_cursor SYS_REFCURSOR;
BEGIN
   OPEN v_cursor FOR
       SELECT employee_id, first_name, last_name, salary, department_id,
              effective_date, end_date
       FROM employee_history
       WHERE employee_id = p_employee_id
         AND p_date BETWEEN effective_date AND end_date;
   
   RETURN v_cursor;
END;
/

-- Usage example
DECLARE
   v_cursor SYS_REFCURSOR;
   v_emp_id NUMBER;
   v_name VARCHAR2(100);
   v_salary NUMBER;
   v_dept NUMBER;
   v_start DATE;
   v_end DATE;
BEGIN
   v_cursor := get_employee_at_date(101, DATE '2023-08-15');
   
   FETCH v_cursor INTO v_emp_id, v_name, v_name, v_salary, v_dept, v_start, v_end;
   
   IF v_cursor%FOUND THEN
       DBMS_OUTPUT.PUT_LINE('Employee ' || v_emp_id || ' on 2023-08-15:');
       DBMS_OUTPUT.PUT_LINE('Salary: ' || v_salary);
       DBMS_OUTPUT.PUT_LINE('Department: ' || v_dept);
   ELSE
       DBMS_OUTPUT.PUT_LINE('No data found for that date');
   END IF;
   
   CLOSE v_cursor;
END;
/

Before Runninginitial state

Before

employees

emp_id	emp_name	dept_id	salary
101	Alice Johnson	10	50,000
102	Bob Smith	20	62,000
103	Carol White	10	55,000
104	Dave Brown	NULL	70,000

departments

dept_id	dept_name
10	Engineering
20	Marketing
30	Finance

Data state before query executes.

Output / Afterresult

After

EMP_NAME	DEPT_NAME
Alice Johnson	Engineering
Bob Smith	Marketing
Carol White	Engineering
Dave Brown	NULL

4 rows — all employees kept, NULL where no matching department

Key Takeaway

Use Oracle Temporal Validity and Flashback Data Archive features to track historical changes and implement temporal queries for time-based data analysis.

REGION	PRODUCT	TOTAL
East	Widget	1000
East	Gadget	1500
East	NULL	2500
West	Widget	800
West	NULL	800
NULL	NULL	3300

STATEMENT	RESULT
ADD phone_number	Column added, NULL for existing rows
MODIFY phone_number	Column size increased to 30 characters
DROP COLUMN phone_number	Column and its data permanently removed

EMP_ID	EMP_NAME	SALARY
101	Alice Johnson	53,000
102	Bob Smith	68,200
103	Carol White	55,000

OBJECT_TYPE	RETURNS_VALUE	CALLABLE_FROM_SQL
FUNCTION	Yes — single value via RETURN	Yes, directly in SELECT
PROCEDURE	No — uses OUT parameters instead	No, only via PL/SQL block or CALL

TRIGGER_TYPE	FIRES
BEFORE, row-level	Once per row, before the row is changed
AFTER, row-level	Once per row, after the row is changed
BEFORE, statement-level	Once before the statement runs
AFTER, statement-level	Once after the statement completes

SALES_MONTH	AMOUNT	PREV_AMOUNT	NEXT_AMOUNT
2026-01	10000	NULL	12000
2026-02	12000	10000	9000
2026-03	9000	12000	NULL