Part 1 of 6 · 1 min read
Why this page follows isolation levels
In this module you will
- State what serializable isolation promises and why weak levels are hard to reason about.
- Explain actual serial execution, and the conditions that make it practical.
- Describe two-phase locking, shared and exclusive locks, deadlock and predicate locks.
- Explain how serializable snapshot isolation detects trouble at commit time.
- Choose between pessimistic and optimistic concurrency control for a workload.
Module 2 listed bugs that weak levels allow and patches for each. Patching one bug at a time is tiring and risky. The book "Designing Data-Intensive Applications" gives the answer researchers have repeated since the 1970s: use serializable isolation. This page shows how databases build it and what it costs. The numbers are made up to show the idea, unless the text names a book.