Understand how to judge when concurrent transactions are correct, not just fast.
This book examines what makes a schedule of database operations legal, and why some forms of concurrency are safer than others. It explains a new semantic idea called independence and how it helps distinguish between serializable, correct, and racing-free schedules.
Readers will see why traditional serializability can be too restrictive for real systems, and how a middle ground—with an easy membership test—helps balance safety and performance. The discussion uses practical examples and a clear, accessible line of reasoning that stays close to the core concepts of integrity constraints and how transactions interact.
- What correctness means in practice: preserving database integrity, isolating transactions, and avoiding racing
- How independence and semantical consistency differ from purely syntactic tests
- The idea of weak serializability and its implications for real-world scheduling
- Conditions under which a middle class of schedules can be tested efficiently
Ideal for readers of database theory and system design who want a grounded, concept-driven view of how concurrency affects correctness.