Multi Region Cloud Architecture for Data Residency Compliance
summary
Atlassian builds a multi-region model with a cloud provisioner, catalog and tenant-contact service that pin a tenant to a realm; migration is a first-class flow handled by a tenant-placement job, a per-tenant looper and a media service using DynamoDB with lazy GSI sharding. The platform manages ~600 services, deploys ~180 per region, supports moving ~10 TB and ~10 million issues with under 8 hours downtime, automates deployment via Spinnaker and achieves ~400% faster migrations while avoiding noisy-neighbor WCU limits.
AI-written summary · about
chapters (9)
- 0:00Introduction to data residency
- 1:09Defining data residency
- 4:08Atlassian's regional expansion
- 6:04Architecting for regions
- 11:15Modeling the data plane
- 13:59Infrastructure and tooling
- 16:55Validation and testing
- 20:03Migration strategies
- 23:54Scaling data movement
related talks
33mGetting Cloud Migration Right
51mAchieving Resiliency With A Multi-Cloud Or Multi-Region Approach
58m’s journey to cloud-native architecture
35mMigrating Your Old Server Products to Be Stateless Cloud Services
43mGetting to a Multi-Cloud Deployment - A Case Study of
21mMulti-Region, Environment Isolated Ray Environments in the Notion Way
44mMulti‑region architectures on AWS
29mScaling Spinnaker for multi-region multi-cloud SaaS deployments