Productionise any Application in Kubernetes Using Sidecar Containers
summary
Demonstrates adding sidecar containers to a running app to fix failures without changing code: deploys an HAProxy sidecar (exposes Prometheus metrics scraped by Prometheus operator), adds a log adapter that tails app JSON logs to stdout for ELK, and inserts an ambassador using go-breaker to return canned responses. A synthetic load with hey reveals 500s caused by “random number divisible by three”; the circuit breaker lowers API load and drops errors from ~25% toward the 10% SLO, so the approach works.
AI-written summary · about
chapters (13)
- 0:00Start
- 4:57Human Needs
- 5:29Reliability
- 8:05ConfigMaps and Secrets
- 8:15Kubernetes Services
- 8:33Kubernetes Ingress
- 8:57Custom Resource Definitions and Operator Pattern
- 10:48API Initial Version
- 11:17Frontend Initial Version
- 11:33Patterns
- 14:28Log Adapter
- 17:08Circuit Breaker Ambassador
- 24:40Service Meshes
related talks
27mPragmatic Pod Patterns: Leveraging sidecar containers in Kubernetes
27mOperationalizing Kubernetes Sidecars in Production at Salesforce
25mRelease the Kraken: Bring Sidecar Containers to Next Level
56mHow to get a grip on your microservices system using a service-mesh
29mThe Great Sidecar Debate
35m10 Weird Ways to Blow Up Your Kubernetes
44mFrom single container to applications at scale
29mAccelerate AppDev with KIK (Kubernetes-Istio-Knative)