Define the CI process • Continuous build triggered for a PR merge • Continuous rolling pipeline triggered by a new build • Continuous deployment to PPE triggered by a green rolling • Deployment auto rollback based on health signals • Dashboards showing the CI pipeline status Build process • Reduce build time • Reduce unit test time • Enable continuous integration in the build pipeline Rolling improvement • Use build artifacts instead of builds in Cloud Vault • Reduce rolling run time Deploy to PPE • Any rolling green build will be pushed to PPE • Critical to have service heath monitoring mechanism in place to prevent bad builds deployed to PPE • Add runners/alerts support in PPE • Treat PPE incidents the same as other environments (except no night call to service team) • Support fast rollback Service health monitoring • Ev2 based on ICM/Azure Monitor/REST: https://ev2docs.azure.net/features/safety/healthcheck/geneva/intro.html • Service Fabric service health model: Service can report health on replica through API: https://docs.microsoft.com/en-us/azure/service-fabric/service-fabricdiagnostics-how-to-report-and-check-service-health Controlled deployment order • Services still have dependencies among each other which can cause issues in PPE • Need a mechanism to define preconditions to block certain service deployments Dashboards • Builds • Rolling test runs • Deployments • PPE build version vs. recent green builds PPE Environment • One environment to host Media Services + LVA?