Navigating Distributed Databases: Challenges & Opportunities
summary
CockroachDB claims Postgres-compatible SQL on a distributed key‑value core, using Raft ranges to give a single-node illusion while allowing multi-primary writes across nodes; teams describe moving banks, gaming and DoorDash workloads off sharded Postgres to one database, onboarding top US banks and a large Indian private bank. Engineering choices include Spanner-inspired consensus, routing around partial network partitions, regional‑by‑row data placement for GDPR, and heavy reliability work: metamorphic testing plus randomized roach tests, Jepsen, spot‑instance chaos and TPCC/YCSB benchmarks.
AI-written summary · about
chapters (12)
- 0:00CockroachDB introduction
- 1:56Adopting CockroachDB struggles
- 3:42SQL vs key-value choice
- 5:31Typical customer problems
- 8:41Multi-primary architecture
- 10:08Spanner and distributed design
- 12:21Manual sharding challenges
- 15:11Production distributed challenges
- 18:04Testing methodologies
- 20:53CAP theorem approach
- 21:36GDPR compliance
- 23:04Network topologies
related talks
42mThe hows and whys of a distributed SQL database
48mBuilding mission-critical applications with CockroachDB
01h 11mFearless global transactions with
52mCockroachDB
49mPostgreSQL’s Journey to Distributed Databases
15mJourney from PostgreSQL to CockroachDB: Scaling Cisco AI Defense
31mFuture-proofing operational data at scale
37mRunning a Cloud-Native SQL Database on Mesos