How we build Data Residency for Atlassian Cloud

This post details the initial steps in Atlassian's evolution towards a multi-region architecture for its cloud products, driven by the need for data residency and latency optimization. It outlines the problems encountered (latency, data storage restrictions, cost efficiency), defines guiding principles (realm definition, service sharding, canonical shard selection, realm migration), and introduces key centralized systems (Cloud Provisioner, Catalogue Service, Tenant Context Service) that orchestrate data plane and control plane operations for multi-region deployments and data migration. The post also discusses the initial migration of Jira and Confluence to a new AWS region and the performance improvements observed.