Concurrency

• Concurrency is achieved when transactions are not executed serially. Their instructions are interleaved with each other.
• This is good because we can stick instructions working on the same resources together, saving the amount of computation time and disk reads needed.

Concurrency Control

• A schedule must be serialisable, recoverable, and preferably cascaedeless.
• This is achieved by a serial execution, but that is not desirable.
• Concurrency-control schemes perform a tradeoff between concurrency and incurred overhead.