API Gateways - The Good, the Bad and the Ugly
summary
Shows a hand-built Apache Camel API gateway and load-tests it with Docker Compose and JMeter, revealing 50 users → 4–6s average, 100 users hits typical client timeouts, and 150 users >60s in some runs. Demonstrates a delete-with-body failure across gateway vs backend, introduces testcontainers + Spring Cloud Gateway for local integration, then compares mitigations: Istio/Envoy service mesh, sidecars, circuit breakers, canaries, and chaos/ephemeral testing.
AI-written summary · about
chapters (12)
- 0:00Introduction
- 7:29Types of API gateways
- 15:04API gateway risks
- 17:12Gateway loopback demo
- 23:10Service mesh solutions
- 28:28Service mesh benefits
- 29:54Building custom gateways
- 32:23Gateway failure modes
- 32:44REST delete body issue
- 35:14Testing API gateways
- 38:04Chaos engineering
- 40:30Anti-patterns and solutions
related talks
20mAPI Gateway in a Nutshell
26mAPI Gateway, A Success Story
29mTaming the Traffic: Select...
36m10 Things You Hate About Your API Gateway
51mThe API Gateway is dead! Long live the API Gateway!
28mAPI Gateway: critical component of your digital architecture
51mSingle Entry Point or Single Point of Failure: The Path to a Non-blocking API Gateway
47mSpring Cloud Gateway: Resilience, Security, and Observability