Part 1 of 5 · 1 min read
Why this page comes first
In this module you will
- Explain the gap serializable snapshot isolation closes, compared to plain snapshot isolation.
- Explain how dangerous structures are detected without resorting to plain locking.
- Explain what happens to a transaction once a dangerous structure is detected.
- Compare serializable snapshot isolation's trade-offs against two-phase locking's.
The previous page showed concrete anomalies, write skew and certain phantoms, that plain snapshot isolation allows. This page covers how some databases close that exact gap, achieving true serializability while still keeping most of MVCC's reader-writer non-blocking benefit from two pages ago.