Growing at the Edge: Doubling Traffic While Changing the API Gateway
summary
Service migrates two API gateways under a nine‑month deadline while onboarding new marketplaces and doubling traffic; evaluation compares nine gateways and runs POCs, selecting KrakenD over Spring Cloud Gateway, then deploys via canary releases, endures seven deployments with rollbacks and incidents, adds a missing HEAD method, raises test coverage from ~60% to ~96%, halves CPU per instance and reaches 65k req/s, handles ~700M image uploads/month, tightens logging and tracing, and lowers operational cost.
AI-written summary · about
chapters (15)
- 0:00Introduction
- 3:18Catalan face words
- 4:00Speaker and Adevinta mission
- 6:27Edge team and JAMS service
- 12:54Replacing legacy API gateways
- 15:18Modern API gateway options
- 17:46Choosing KrakenD
- 19:57Doubling traffic while migrating
- 23:11Fetch gateway with canary releases
- 24:15Adevinta production deployments
- 26:46Fetch gateway rollbacks
- 29:58Management API gateway migration
- 32:10Results after replacement
- 35:45Key takeaways
- 38:58Q&A
related talks
32mThe Past, Present and Future of Cloud Native API Gateways
29mThe API Gateway Maturity Matrix: Where Do You Rank?
23mDerisking Migrations of API Backends
31mOpen Banking APIs and Architecture at Scale
31mIn and Out: A 20-Minute Adventure or How to Rewrite a Bank's API Gateway
19mZero-Downtime by Design: High-Availability Migration Patterns
27mManaging the Complexity of Microservices Deployments - Prithpal Bhogill, Google & Kenny Bastini
18mScaling APIs to Billions: Extreme-Scale Reliability