Atlassian migrated 4 million Jira PostgreSQL databases to Amazon Aurora, aiming to improve reliability and decrease costs. The architecture employs one database per tenant, enhancing isolation and scalability necessary for Atlassian's scale. The migration project, estimated to last a few months, sought to minimize tenant downtime and costs. By using AWS Step Functions and feature flags, the team efficiently managed the transition, optimizing database endpoints quickly. The move to Aurora will leverage its SLA, improve scalability, and provide cost efficiencies as the team maintains a balanced load across instances.
Atlassian migrated over 4 million Jira PostgreSQL databases to Amazon Aurora to improve reliability and reduce costs, developing custom orchestration tools for the transition.
The architecture of utilizing one database per tenant helps Atlassian maximize isolation, scalability, and operational control, crucial for managing their vast user base.
Migrating to Aurora aimed to leverage better SLA (99.99%), increase autoscaling capabilities, and achieve cost savings while minimizing tenant downtime during the process.
Implementing AWS Step Functions and feature flags was essential for a successful migration, enabling immediate updates to tenants' database endpoints with minimal disruption.
Collection
[
|
...
]