How we catch and mitigate performance regressions at scale in Jira Cloud

This post details the implementation of a per-tenant, per-endpoint performance regression detection system for Jira Cloud. It explains the challenges of performance variation across tenants and the limitations of aggregated monitoring. The system uses statistical process control (SPC) to detect regressions by comparing current performance to historical baselines for each tenant and endpoint. It integrates with Rovo Dev CLI to automate root cause analysis by correlating performance data with feature flag evaluations and source code changes, as demonstrated by a real-world regression involving synchronous network calls to the Tenant Context Service.