Writing Reliable, Scalable, Fault Oblivious Code On K8s the Easy Way
summary
Demo code wires a front end -> order API -> Kafka -> archive -> scheduled loyalty job architecture running on Kubernetes, using Dapr sidecar and over 100 Dapr components. Chaos injected via Azure Chaos Studio (chaos-mesh) adds network delay; non-Dapr calls return 500s while Dapr applies a resiliency spec (500ms timeout, retries, circuit breaker at 20) to hide failures. Distributed lock (Redis lease lock) serializes a loyalty job running every 15s so the locked run reliably produces 20 orders versus the no‑lock run producing inconsistent counts (30s+).
AI-written summary · about
chapters (6)
- 0:00Start
- 0:52Developer challenges
- 2:49DAPR COMPONENTS
- 3:53End-to-end solution
- 9:19Dapr Resiliency: Service Invocation
- 30:34QUESTIONS?
related talks
30mFailure Is Not an Option: Durable Execution + Dapr = 🚀
49mStart Building Distributed Applications with Ease Using Building Block APIs
21mShaping next generations of architectures with Dapr
34mAutoscaling Made Easy with Kubernetes Event-Driven Autoscaling (KEDA)
01h 00mEvent-driven computing with Kubernetes
55mStop Caring about Kubernetes with Azure Container Apps & Dapr
31mDapr + Score: Mixing the Perfect Cocktail for an Enhanced Developer Experience
46mAvoid mastering Kubernetes for local dev. with Dapr and Testcontainers