Achieving Resiliency With A Multi-Cloud Or Multi-Region Approach
summary
A single-region Spring Boot app is rebuilt as a active-active system: identical services run in North Europe and South Central US on Azure Spring Apps, Azure Cache for Redis Enterprise uses conflict-free replication, and Azure Front Door performs health checks every 5s and routes traffic. The stack is deployed and torn down with Terraform for game-day testing; deleting an entire region proves traffic shifts to the other region and latency improves (example: ~229 ms down to ~150 ms).
AI-written summary · about
chapters (18)
- 0:00Introduction and speaker background
- 3:09Zones regions and low latency
- 4:55Resiliency vs robustness
- 8:23Terraform and infrastructure setup
- 11:11Leader follower pairs in a region
- 13:44Active passive limitations
- 16:52Active active advantages
- 21:51Active active approaches
- 25:02Demo app and build packs
- 29:55Why use build packs
- 33:23Centralized build processes
- 37:38ARM64 and cost savings
- 40:07Cost per transaction and latency
- 42:12Multi region deployment
- 45:43Active active without code changes
- 46:25Front door and health checks
- 49:12Game day and region deletion
- 50:56Summary and takeaways
related talks
52mMy Children Will Never Deploy Active-Passive
22mCloud-Native Apps, recovery ready!
25mChaos Engineering for Cloud native Apps
22mCloud-Native Apps using Azure Cosmos DB, ready for recovery!
39mExperiences and Lessons Learned Building a Multi-Cloud Spring Boot Backend
29mMulti Region Cloud Architecture for Data Residency Compliance
27mAccelerate Spring Apps to Cloud at Scale—Discussion with Azure Spring Cloud Customers
39mMaking Production as Smooth as Development with Azure Spring Apps