Transactions: myths, surprises and opportunities
summary
Asserts that transactions are resurging with practical, scalable implementations: MVCC snapshot isolation (Postgres, MySQL) and optimistic conflict checks (Postgres SSI) avoid classic 2PL bottlenecks, while in-memory serial execution (H-Store/VoltDB) trades latency for simplicity. Demonstrates concrete anomalies—read skew and write skew—and shows how defaults (many DBs use read committed) silently allow them. Explains distributed costs: atomic commit/consensus (2PC, ZooKeeper) is expensive, so systems use compensating transactions or aim for causal consistency as a middle ground.
AI-written summary · about
chapters (4)
related talks
45mTransactions and Concurrency Control Patterns
41mConcurrency, Scalability and Transactions -- Myths and Surprises
52mTransactions: Myths, Surprises and Opportunities
43mBeing Transactional in 2019 — What Does That Actually Mean?
28mUnderstanding Transaction Isolation for App Developers
47mBeing transactional in 2019 - what does this actually mean?
17mTransactionalism Explained
35mDon't Give Up on Serializability Just Yet