Database Drivers: Performance Perspectives
summary
Database drivers matter for real behavior not buzzwords: expect concrete guidance on caching prepared statements (avoid preparing one-off startup queries; database evicts with LRU-ish policies), on retries (only idempotent operations, use backoff and avoid retry storms, only idempotent retries), and on timeouts (recommendation: server timeouts < client timeouts + network latency). A concrete bug describes a Cassandra Python driver bug where reusing request IDs after client timeouts caused mismatched responses and account mixups. Also watch measurements: 20 MB/s wire can hide 1 MB/s goodput.
AI-written summary · about
chapters (9)
- 0:00Introduction to drivers
- 2:33Client-server relationship
- 6:09Timeouts and retries
- 9:35Topology awareness
- 11:40Caching in drivers
- 14:54Retry strategies
- 17:44Paging results
- 18:47Concurrency in drivers
- 20:09Summary and Q&A
related talks
01h 05mHigh-Performance JDBC
40mLatency, Throughput & Fault Tolerance: Building a Modern Database
58mWhy Databases Cry At Night
37mPopular Apache Cassandra Problems & How to Fix 'Em
33mCassandra Performance Tuning Like You've Been Doing It for Ten Years
26mA deep dive and comparison of Python drivers for Cassandra and Scylla
52mJava Persistence and Hibernate Tips that can boost up your application performance
39m10 Things Developers should know about Databases