Cross-Region Replication Options with FoundationDB
summary
FoundationDB 6.0 adds multi‑region support that keeps commits local while protecting against region failure by doing synchronous logs across availability zones and asynchronous physical storage replication between regions. A new log router pulls each mutation exactly once so cross‑region traffic is minimized; cluster keeps four storage replicas and can fail over automatically in seconds (upper bound ~30s). Current limits: two‑region support only, throughput degrades while a region is offline and operators must monitor replication lag; a manual data‑loss failover option also exists.
AI-written summary · about
chapters (10)
- 0:00Intro
- 1:08Motivation
- 1:34Disaster Recovery
- 3:51Policy Based Replication
- 7:09Multi-Region Tradeoffs
- 7:45Region Configuration
- 8:01Resilience
- 10:09Performance
- 15:02Foundation DB 6.0
- 18:02Monitoring Data Lag
related talks
31mDeep dive on cross-Region DR with Amazon Aurora global databases
34mFoundationDB, the Black Knight
11mConsistent Caching in FoundationDB
18mLeveraging Multiple Regions to Improve Site Reliability:
24mRetrofitting a Foundation for CouchDB
24mHow We Saved 5x Migrating from Aurora to FoundationDB
27mAvailability and Performance Tradeoffs in Global Database Deployments
16mFoundationDB and Node.js: Year in Production