Network Observability with Envoy
summary
Envoy emits listener, cluster and request metrics that are actionable: monitor connection totals and SSL verify failures for mTLS, watch upstream retry and timeout counts, and alert on outlier ejections and authorization failures. Use stats tags to convert metric name parts into searchable tags rather than long raw names. Export via StatsD/DogStatsD or Prometheus (note Prometheus histograms in Envoy 1.10), secure the admin metrics endpoint. Decode gRPC with the HTTP/1 bridge filter for gRPC status and send tracing via a collector.
AI-written summary · about
chapters (19)
- 0:00Start
- 6:35Observability
- 7:39Internal and external instrumentation
- 9:47Upstream Terminology
- 12:29Envoy Prometheus Metrics Metrics
- 16:23Metrics types Prometheus
- 18:35Key service metrics
- 19:32Total connections Grafana
- 20:24Key diagnostics metrics
- 23:26Request Errors
- 26:00gRPC Bridge Filter Configuration
- 27:40gRPC Error
- 28:06Key metrics Cluster
- 28:35Retries
- 28:58Timeouts
- 29:25Outlier Ejection
- 32:19What is tracing? Tracing
- 32:41Tracing Cluster
- 35:04Logging
related talks
33mGet to know Envoy - the modern proxy for cloud-native infrastructure - Baptiste Collard
38mEnvoy Deep Dive
23mUsing Envoy as an Egress Proxy for TLS Enabled Traffic
33mIntro: Envoy
27mEnvoy Proxy: Evolved for Serving LLMs
34mIntro: Envoy
27mEnvoy - Overview and Update: Senior Envoy Maintainer
28mDeploying hybrid topologies with k8s & Envoy: a look at service discovery