Course contentsShow
← Cloud and Infrastructure
Lesson 1633 of 2,98933. Service Mesh FeaturesFree lesson

Golden Signals in Service Mesh

Explore latency, traffic, errors, and saturation metrics exposed by service mesh proxies.

Golden Signals in Service Mesh

What you'll learn: How service mesh proxies automatically expose the four key metrics (latency, traffic, errors, saturation) that reveal service health at a glance.

The Four Golden Signals

Service meshes excel at observability because their sidecar proxies sit in the request path and automatically collect critical metrics. The "Golden Signals" framework—popularized by Google's SRE practices—identifies four vital health indicators:

1. Latency

Time taken to service requests. Service mesh proxies measure request duration from arrival to response, capturing percentiles (p50, p95, p99) that reveal user experience. Slow responses often indicate downstream issues or resource constraints.

2. Traffic

Volume of demand on your service. Proxies count requests per second, helping you understand load patterns and capacity needs. Think of it as measuring how many customers are entering your store at any given moment.

3. Errors

Rate of failed requests. The mesh tracks HTTP status codes (4xx, 5xx) and connection failures, instantly revealing when something breaks. An error spike is your service raising its hand for help.

4. Saturation

How "full" your service is—CPU, memory, connections. While proxies primarily measure connection pool utilization and queue depths, they complement infrastructure metrics to show when you're approaching resource limits.

Why Service Mesh Makes This Easy

Without a mesh, you'd instrument each service individually. With service mesh proxies automatically collecting these metrics, you get uniform, comprehensive observability across all services from day one—no code changes required. The proxies expose these as Prometheus metrics by default, ready for dashboards and alerts.

Key Takeaway: Service mesh proxies automatically collect the four Golden Signals (latency, traffic, errors, saturation), giving you instant, standardized visibility into every service's health without manual instrumentation.