Slide 4 of 12

Istio integration

Istio integration

The Istio integration provides service mesh monitoring for Kubernetes environments running Istio.

What it’s forMonitoring service-to-service traffic, latency, and mesh health
Who uses itPlatform engineers, SREs managing Kubernetes with service mesh
Under the hoodCollects metrics from Istio control plane and Envoy sidecars

Metrics collected

  • Traffic — Request volume by service pair
  • Latency — P50, P90, P99 by service
  • Errors — HTTP status codes, failures
  • Envoy — Sidecar proxy metrics

Trade-offs

Best for: Kubernetes environments running Istio service mesh

ProsCons
Pre-built dashboards: mesh overview, servicesIstio-specific concepts to learn
Pre-built alerts: error rates, latencyHigh cardinality in large meshes
Service-to-service visibilityRequires Istio to be deployed

Documentation

View the full documentation. Learning path coming soon!

Istio integration

Script

Istio is a service mesh for Kubernetes. It sits between your services and handles traffic routing, security, and observability. The integration gives you visibility into this mesh layer.

Traffic metrics show who’s talking to whom, with request volumes between services broken down by source and destination. Latency metrics surface performance at the mesh level. You can see exactly how long service-to-service calls are taking. Error rates and HTTP status codes identify failing interactions.

The integration also exposes Envoy metrics. Envoy is the sidecar proxy that Istio deploys alongside each service. It’s the component actually handling traffic. Understanding Envoy behavior helps you troubleshoot mesh issues.

If you’re running Istio, this visibility is essential. Service meshes add complexity, and without observability, that complexity becomes unmanageable.

One thing to watch: cardinality. Large meshes with many services generate lots of time series because metrics are labeled by both source and destination service. Consider aggregation rules to keep cardinality manageable.

The integration pairs beautifully with distributed tracing. When you see a latency spike in mesh metrics, you can drill into traces to find exactly which service is slow.

This is advanced infrastructure monitoring, appropriate after you’ve got core Kubernetes observability working.