Leveraging Multiple Regions to Improve Site Reliability:
summary
Jet moves from one Azure region to multiple by replicating a single event store replication as the source of truth, projecting caches and indexes locally, and shifting from active/passive to an active‑active stateless pattern where feasible. A chain‑replication service does asynchronous cross‑region sync, weekly traffic cutovers and monthly storage failovers reveal missing dependencies, and a homebuilt state‑machine failover tool replaces brittle scripts to prevent dual‑master divergence. Three regions cut required spare capacity roughly in half; the approach is effective but requires ongoing exercises and team remediation.
AI-written summary · about
chapters (7)
- 0:00Intro
- 2:57Why did we go multi-region?
- 5:03Don't replicate every data store
- 8:57Not every system needs to be Active/Active
- 10:19Three is cheaper than two
- 11:45Practice. Practice, Practice
- 13:35Failover Automation Needs to Scale
related talks
19mPreparing for disaster: going Multi-Region
59mBuilding Resilient Applications In Microsoft Azure
52mMy Children Will Never Deploy Active-Passive
44mAvailability, Latency and Cost: Withstanding Outages
20mManage Scaling, While Keeping Costs Under Control. Challenge Accepted.
48mReplication as a First-Class Citizen: The Art of Global Resilience
01h 00mBuilding Resilient Applications In Microsoft Azure
18mScaling APIs to Billions: Extreme-Scale Reliability