1. Flat Transaction Model: Simplest model, entire transaction as one unit, follows ACID. Example: BEGIN; Update A; Update B; COMMIT. Failure → full rollback.
2. Nested Transaction Model: Transaction divided into sub-transactions. Structure: T { T1, T2, T3 }. Advantages: parallelism, fault isolation.
3. Distributed Transaction Model: Executes across multiple sites, uses Two-Phase Commit (2PC). Example: Bank transfer (debit site A, credit site B).
4. Chained Transaction Model: Series of smaller independent commits, improves recovery.
5. Workflow Transaction Model: Tasks follow business process, used in e-commerce and supply chains.
Conclusion: Transaction models improve concurrency, reliability, and distributed execution.
Flat Transaction: Single execution unit: BEGIN → Execute → COMMIT/ABORT. Example: Update account balance; failure causes full rollback.
Nested Transaction: Main transaction contains sub-transactions (children). Example: Online shopping: Parent = Place order; Children = Payment, Inventory, Delivery.
| Flat | Nested |
|---|---|
| Single unit | Multiple sub-units |
| Full rollback | Partial rollback |
| Less parallel | More parallel |
Serializability: Concurrent execution produces same result as some serial execution.
Conflict Serializability: Uses precedence graph; read-write/write-write conflicts must be acyclic.
View Serializability: Schedules appear identical to serial schedule (same initial reads, final writes).
Importance: Prevents inconsistency, maintains correctness, improves concurrency.
Ensures all sites commit or abort together → atomicity in distributed transactions.
Example: Bank transfer: debit success, credit failure → without protocol data becomes inconsistent.
Protocols: Two-Phase Commit (2PC), Three-Phase Commit (3PC).
Coordinator failure: Participants wait, recover from log.
Participant failure: Recover from stable storage, use undo/redo.
Communication failure: Timeout mechanism triggers recovery.
Shared Memory: CPUs share memory and disk. Fast communication but limited scalability.
Shared Disk: Each CPU has private memory, shared disk. High availability but disk contention.
Shared Nothing: Each node has CPU+memory+disk. Highly scalable, fault tolerant, complex communication.
Goals: Improve speed, handle large data, load balancing, fault tolerance, high throughput.
Ensures concurrent execution = serial execution order. Benefits: consistency, correctness, no lost updates.
Techniques: Locking (2PL), Timestamp ordering, Optimistic validation.
Integrates multiple autonomous databases under global schema.
Components: Global schema, local schema, data translator. Advantages: Autonomy, scalability. Disadvantages: Complex integration, query optimization overhead.
Database that supports mobile users with wireless access, replication, and synchronization.
Characteristics: Portability, intermittent connectivity, data caching.
Applications: Banking, healthcare, navigation, e-commerce.
Techniques:
Benefits: Consistency, deadlock control, isolation, improved throughput.
✔ Complete 15-mark answers for MAKAUT PEC-IT601B (2024–25).
✔ Covers transaction models (flat, nested, distributed, chained, workflow), 2PC (need, failure response, disadvantages), parallel architectures (shared memory/disk/nothing), serializability, multidatabase, mobile database, concurrency control.
✔ Answers structured with definitions, examples, tables, and diagrams per MAKAUT exam pattern.