Course contentsShow
← System Design
Lesson 1205 of 1,91930. Observability: MetricsFree lesson

OpenTelemetry Metrics SDK

Learn how OpenTelemetry standardizes metrics instrumentation and provides vendor-neutral metrics collection.

OpenTelemetry Metrics SDK

What You'll Learn

You'll discover how OpenTelemetry provides a vendor-neutral, standardized approach to collecting and exporting metrics from your distributed systems.

The Core Problem

You've learned about Prometheus, StatsD, CloudWatch, and other metrics platforms—each with its own SDK, data model, and instrumentation patterns. Switching vendors traditionally meant rewriting all your instrumentation code. This creates vendor lock-in and engineering overhead.

What is OpenTelemetry Metrics SDK?

OpenTelemetry (OTel) is an open standard that unifies metrics, traces, and logs under one instrumentation layer. The Metrics SDK specifically handles creating, collecting, and exporting metrics using a vendor-agnostic API.

Think of it like USB-C for metrics: you instrument your code once using OpenTelemetry's standard interfaces, then configure exporters to send data to any backend (Prometheus, Grafana, CloudWatch, etc.) without changing your application code.

How It Works

OpenTelemetry provides standard metric instruments (counters, gauges, histograms) through a unified API. You instrument your code using these primitives:

// Conceptual illustration
meter = otel.getMeter("my-service")
requestCounter = meter.createCounter("http_requests_total")
requestCounter.add(1, labels={"method": "GET", "status": "200"})

The SDK handles aggregation and periodically exports to configured backends via exporters (OTLP protocol, Prometheus scrape endpoint, etc.).

Key Benefits

  • Vendor portability: Switch backends without reinstrumentation
  • Standardized naming: Consistent metric semantics across tools
  • Multi-backend support: Export to multiple platforms simultaneously
  • Rich context: Built-in correlation with traces and logs

Key Takeaway: OpenTelemetry Metrics SDK standardizes metrics instrumentation with vendor-neutral APIs and exporters, letting you instrument once and send data anywhere without code changes.