Sandbox environments for development

Transcript

1. Sandbox Environments Engineering Team
2. Problems - Our Staging is having many problems: - It is shared among all teams. - One broken changes could bring down the whole staging environment. - Staging is not actually a real staging.
3. Staging Feature A (owned by team X) Feature B (owned by team Y) Feature C (rejected by BA, but not being cleaned up) Half of feature D (owned by team Z) Some random test codes (from unknown members) Blah blah ...
5. Problems - Staging and Production diverses more and more - Testing on Staging doesn’t reflect the production environment. - Teams step on each other’s feet. - Staging is not reliable for business uses (demo / present / walkthrough, etc.) - Inconvenient deployment flow
6. Solutions - Each team should have their own staging sandbox - There should be a pre-production environment
7. Sandbox Engineering Sandbox Engagement Sandbox Partnership Aon sandbox/engineering-team sandbox/engagement-team sandbox/partnership-team-aon Sandbox Partnership AMP sandbox/partnership-team-amp Staging development Production master Feature A Feature B Feature C Smoke tests Manual tests Demo Present
8. Sandbox isolation - Each sandbox is isolated. - They have their own: - Web endpoint - Web server, worker servers, rpc server and OCI server - Database (minimal version, only 2 mb) - Memcache server - Redis server - We don’t need to wait for CI
9. Demo
10. Migration plan - Please register your team with Engineering team. - We’ll test the flow on some teams first (1 or 2 week). - Then, each team will manage their own sandboxes. - After that, the pre-production staging will be reseted to “development” branch. - Finally, a new pre-production cluster will be created.
11. Q & A