1️⃣ Partial Booking → Temporary Hold with Expiry

• Problem: In booking systems (e.g., airline tickets, hotel rooms), multiple users may try to book the same resource simultaneously.

• Solution: Use Redis to place a temporary hold on the item.

  • Store the booked item in Redis with a TTL (Time To Live).
  • If the user completes the booking, sync with the DB.
  • If the TTL expires, the hold is automatically released.

Example:

User selects seat 12A → Redis key: "seat:12A" = "held" (expires in 5 min)

✅ Pros:

• Very fast (Redis is in-memory).
• Automatically cleans up expired holds.

---

2️⃣ Rollback → Auto-expiry + Compensation (not XA transaction)

• Problem: Distributed transactions (DB + Redis) cannot always use XA transactions (two-phase commit) due to complexity.

• Solution: Use auto-expiry in Redis + compensation logic in the DB:

  • Redis key expires → system triggers a rollback or compensating action in the DB.
  • Example: Refund reserved amount, release inventory.

Example Flow:

User holds item → TTL expires → Job detects expired hold → Release DB lock → Refund amount

✅ Pros:

• Avoids complex distributed transactions.
• Handles failures gracefully.

---

3️⃣ Concurrency → Row Lock or Optimistic Versioning

• Problem: Multiple users may try to update the same row in the DB concurrently.

• Strategies:

  1. Row-level locking (pessimistic): Lock the DB row when updating.

     • Pros: No conflicts.
     • Cons: May block other transactions → less concurrency.

  2. Optimistic versioning: Use a version column to detect conflicts.

     • Pros: High concurrency, no blocking.
     • Cons: Need retry logic if version mismatch occurs.

Example:

UPDATE bookings
SET status='confirmed', version=version+1
WHERE id=123 AND version=2;
-- fails if version changed

---

4️⃣ System Crash Recovery → Cleanup via Cron/Job

• Problem: System may crash after a hold is placed in Redis but before syncing with DB.

• Solution: Use a scheduled cleanup job (cron or background worker):

  • Scan Redis for expired or orphaned keys.
  • Release any associated locks or resources in the DB.

Example Flow:

Cron job runs every minute → Checks Redis for expired holds → Releases DB rows/resources → Logs cleanup

✅ Pros:

• Ensures system consistency even after crashes.
• Prevents stale locks or ghost bookings.

---

Putting it all together: Workflow

1. User requests booking → place temporary hold in Redis with TTL.
2. User confirms → update DB with optimistic versioning or row lock.
3. If user cancels or TTL expires → compensation logic releases DB resources.
4. Cron/cleanup job ensures system crash recovery.

---