Migrating to Open Feature at Scale: 0 to Billions
Talkat OpenFeature Summit NA 2024 · on Tuesday, 2024-11-12· at 10:40· Salt Palace | Level 2 | 250 D·
summary
eBay replaces 25 legacy config APIs with an OpenFeature facade and an internal provider, authors a Java SDK, and folds flags into the existing Touchstone experimentation lifecycle. Migration runs three quarterly pilot waves with 21 teams, 300+ migration jobs and dedicated velocity budgets, yielding 1 minute change propagation, 3 ms evaluation latency, trillions of flag evaluations per day and ~2,500 simultaneous experiments. Practical wins: unified targeting (including CDP segments and internal beta audiences), error monitoring, and documented onboarding that cuts review-bound delays and enables massive rollout.
AI-written summary · about
chapters (12)
- 0:00Open Feature at eBay
- 1:04Why Open Feature
- 3:10Velocity program’s role
- 6:42Legacy flag problems
- 9:37Touchstone platform
- 11:22Feature flag requirements
- 12:45Onboarding developers
- 13:48Pilot phases and MVP
- 16:38Migration strategy
- 18:25Scale and performance gains
- 19:29Future developer workflow
- 21:58Q&A
related talks
38mFun with Flags: How OpenFeature Solves Your Feature Flag Headaches
43mFeature Flags with OpenFeature: Faster, More Collaborative Development
32mOpenFeature Update From the Maintainers
26mBuild Progressive Delivery and Safety Into Your Apps with Feature Flags
43mOpenFeature -- one flag to rule them all
25mType-safe Feature Flagging in OpenFeature: Lessons Learned From Using Feature Flags at Google
05mFeature Flags
16mAdvanced Feature Flagging: It's All About The Data