Question

According to the conflicts and failure in transactions processing, the need of concurrency control and recovery appeared.
Discuss the differences between concurrency control and recovery in terms of:
The purposes.
The algorithms.
The problems.

thank you.

Answer 1

Differences between concurrency control and recovery in terms of the purposes,algorithms and problems are as follows:

	Concurrency Control	Recovery
Purpose	There are multiple transactions happening over a single shared database which might lead to integrity & consistency problems. To avoid such problems with the shared database, we use concurrency control. The purpose is to make every individual user feel the consistency of the system even though it is been used by multiple user at the same time. To ensure serializability To apply isolation using mutual exclusion between conflicting transactions.	Recovery process is needed by the systems to recover from the unforeseen failures if any. i.e to make the system fault tolerant. The purpose here is to undo the changes done by aborted/incomplete transactions and only consider the transactions that have been committed/completed successfully. The purpose is coping with all kinds of failure-transaction, system or media
Algorithms	Lock based Graph based Timestamp based Validation based Multiversion schemes	STEAL and NO-FORCE policy Logging Write Ahead Logging (WAL)
Problems	Updates might be lost if multiple transactions select same rows and try to update it. Dirty reads can take place. Second transaction might try to access the same row multiple times with different value each time.	Backups are needed to recover from media failure which is time consuming and requires down time. Backup frequency should be decided carefully. If transaction log itself are damaged then restoring the database from the last backup becomes impossible.