OpenTelemetry SDKs

SDK Overview

OpenTelemetry SDKs provide the API and implementation for producing telemetry. Each language has its own SDK package. The SDK configures exporters, processors, samplers, and resources that define your telemetry pipeline.

Resource Configuration

A Resource describes the entity producing telemetry. It includes service.name, service.version, deployment.environment, and custom attributes. Resources are attached to every signal (traces, metrics, logs) from the service.

from opentelemetry.sdk.resources import Resource, SERVICE_NAME

resource = Resource.create({
    SERVICE_NAME: "payment-service",
    "service.version": "2.3.1",
    "deployment.environment": "production",
    "team": "payments",
})

Meter Setup

The Meter produces metrics. Configure the Meter with a name (usually service.name), version, and resource. The Meter creates counters, histograms, and gauges that are exported via OTLP.

Tracer Setup

The Tracer produces spans. Configure the Tracer with a name (usually service.name) and version. The Tracer creates spans that represent individual operations within a request trace.

Logger Setup

The Logger produces log records. OTel log support is newer and still evolving. Configure the Logger with the same resource as traces and metrics to enable correlation between signals.

Semantic Conventions

Semantic conventions define standard names for attributes across languages and libraries. Using conventions ensures consistency. HTTP, database, messaging, and RPC all have defined conventions.

HTTP:
  http.request.method
  http.response.status_code
  url.full
  server.address

Database:
  db.system
  db.statement
  db.name

Messaging:
  messaging.system
  messaging.destination.name
  messaging.operation.type