Unit 4 — Transaction Management & Concurrency Control

Unit Overview

This unit addresses two fundamental challenges in database systems — ensuring that database operations complete reliably even when failures occur, and ensuring that multiple users can access the database simultaneously without corrupting data. These problems are collectively solved through transaction management, concurrency control protocols, and recovery mechanisms.

In any real-world database system, multiple users execute operations concurrently. A banking system processes thousands of transfers per second, an airline reservation system handles simultaneous bookings, and an e-commerce platform updates inventory while multiple customers checkout. Without proper transaction management and concurrency control, these systems would produce incorrect results — money could disappear, seats could be double-booked, and inventory counts could become negative.

Topics Covered in This Unit

The Concurrency Problem Illustrated

This example demonstrates why uncontrolled concurrency is dangerous. The solution requires either locking A during each transaction or using timestamps to detect and resolve conflicts.

How Topics Build on Each Other

The unit follows a logical progression. Transaction management introduces the concept of transactions and their properties. Transaction states describe the lifecycle. ACID properties define what guarantees a correct transaction system must provide.

Concurrency control then addresses how to achieve isolation when multiple transactions run simultaneously. Serializability provides the theoretical correctness criterion. Lock-based protocols and two-phase locking offer one approach to achieving serializability. Timestamp protocols provide an alternative lock-free approach. Deadlocks are a side effect of locking that must be handled.

Recovery mechanisms ensure durability and atomicity — if the system crashes, log-based recovery and checkpoints restore the database to a consistent state. Shadow paging offers a completely different recovery approach without logs.

Learning Objectives

After completing this unit, you should be able to:

1. Define transactions and explain all four ACID properties with practical examples
2. Identify concurrency problems — dirty reads, lost updates, non-repeatable reads, and phantom reads
3. Test whether a schedule is conflict-serializable using precedence graphs
4. Apply two-phase locking protocol and identify its variants
5. Execute the timestamp-based protocol and apply Thomas's write rule
6. Detect deadlocks using wait-for graphs and apply prevention strategies
7. Perform log-based recovery using undo, redo, and undo-redo algorithms
8. Explain checkpoint mechanisms and shadow paging recovery

These concepts are essential for database administrators, backend developers, and anyone working with systems that require data integrity under concurrent access.