The Busy Platform Engineers Guide to API Gateways
summary
Choosing or migrating an API gateway is a type 1 decision because gateways sit on the business-critical hot path, are sticky, and often incur contract lock-in and high cost; migrating requires a migration plan and re-evaluating requirements. Practical examples: Emissary Ingress uses Kubernetes CRDs for mappings so developers can declare routes; Linkerd integration required a one-line integration; avoid embedding business logic in plugins (Zuul groovy scripts anti-pattern). Enable self-service via pull requests and add extra validation for global security changes.
AI-written summary · about
chapters (22)
- 0:00Intro
- 0:23tl;dr
- 2:36Software engineering is all about decisions
- 3:59Choosing an API gateway is a type 1 decision
- 5:55From Kubernetes to PaaS to...?
- 6:29A quick recap for building platforms
- 7:54Looking back at my dev career
- 9:36The need for a platform control plane emerges
- 10:02The CNCF ecosystem is the foundation
- 10:38A word of caution with platforms
- 11:57Building platforms: What did I learn?
- 13:07API Gateway as a Product
- 18:08Modern cloud native comms stack
- 21:10All-in-one or one-for-all?
- 22:39You can't have good DevX without good UX
- 24:58Self-Service configuration
- 29:18Extra validation when applying global configuration?
- 29:41Interop example: Emissary-ingress & Linkerd
- 32:23API gateway plugins: love 'em/hate 'em
- 35:48Wrapping up
- 36:41Conclusion
- 40:02Outro
related talks
30mA Comprehensive Guide to API Gateways, K8s Gateways, and Service Meshes
38mAPI Gateways & Service Mesh: What’s best for us
27mMaster API Gateways: Boost Your Microservices Strategy
27mEnterprise API management in agile integration
39mAPI Gateway to Service Mesh: Navigating a Changing Landscape
26mSynchronous Communication Patterns: A journey from ESB to APIs [...]
30mIs Zero-Downtime Migration Possible? Moving From Ingress & Sidecars To Gateway API
33mFantastic API Gateways and Where to Find Them