Free Transactions and concurrency lessonsTransactions and concurrency · Stage 2 · Control and recovery · Step 3
Checkpoints and crash recovery
See how checkpoints bound recovery work, and the order of the analysis, redo and undo passes. About 13 minutes.
Why does recovery redo the changes of uncommitted transactions too, before undoing them?
- AIt first restores the exact state at the crash (repeating history), so undo can work on a known state
- BBecause they might have committed
- CBecause redo is faster than undo
Show the answer
It first restores the exact state at the crash (repeating history), so undo can work on a known state. That makes recovery simple and safe to repeat.
4 questions on this idea
Short questions that apply the idea to new cases: trace it, work out a number, put steps in order. A wrong answer explains itself.
Answer them in your browserFree account, no card. Your progress is saved, and the idea comes back in a quick review later.
Sign up free to solve