Scaling Django: Mastering Database Routers with AWS Aurora
summary
OneBlinc moves Django reads to Aurora’s read endpoint and keeps writes on the primary via a simple Django database router that maps DB-for-read to a read-only endpoint and DB-for-write to the writer, forces migrations and locks to the writer, and treats relations as allowed. Replication lag is typically 10–20 ms, reader nodes carry 60–70% of traffic, writer CPU drops to ~15%, and failover/promote completes in under 1 minute; autoscaled readers handle payday peaks and cut the DB bottleneck.
AI-written summary · about
chapters (11)
- 0:00Introduction and speaker intro
- 1:11Company background and team
- 3:25Database overload problem
- 8:09Why choose AWS Aurora
- 11:57Database routers explained
- 15:25Solution idea with read-only endpoint
- 18:30Key implementation insights
- 24:00Cost and optimization tips
- 27:36Lessons learned and scaling challenges
- 31:06Future possibilities and alternatives
- 34:37Q&A and closing
