Jaeger Joins the CNCF: Tracing for Cloud-Native Operations

Introduction

On September 13, 2017, the Cloud Native Computing Foundation voted to accept Jaeger—Uber’s open-source distributed tracing system—into the CNCF Sandbox (then called the Incubating program). This move positioned Jaeger as the community-backed tracing solution for Kubernetes-era microservices, complementing Prometheus metrics and Fluentd logs.

What Jaeger Brings

OpenTracing Native: Implements the OpenTracing APIs, allowing language SDKs (Go, Java, Node.js, Python, C++) to instrument services with minimal boilerplate.
Envoy + Istio Integration: Envoy proxies emit spans automatically, giving Kubernetes clusters trace coverage without modifying every service.
Storage Flexibility: Pluggable backends (Cassandra, Elasticsearch, memory) to match operational needs.
UI for Root Cause: Web console visualizes latency breakdowns, critical path and span timelines across services.

Architecture Overview

Component	Role
Client Libraries	Instrument applications, forward spans to local agents.
Agent	Sidecar or daemon that batches spans and ships to collectors.
Collector	Normalizes spans, performs sampling decisions, writes to storage.
Query + UI	Serves the API/UI for retrieving and visualizing traces.

Kubernetes deployments typically run the agent as a sidecar DaemonSet, ensuring low-latency UDP delivery while keeping configurations centralized.

Installing Jaeger on Kubernetes

The community published Helm charts and jaeger-operator manifests with:

kubectl create namespace observability
helm install jaeger jaegertracing/jaeger --namespace observability
Configure sampling via ConfigMaps; enable ingress for the query UI.

Operators can pick all-in-one (dev/testing) or production templates (collectors + storage clusters).

Impact of CNCF Sponsorship

Governance & Neutrality: CNCF steering encourages cross-vendor contributions, ensuring Jaeger remains interoperable with Envoy, Istio, Linkerd and other projects.
Ecosystem Bridges: Prometheus exporters report tracing runtime health; Fluentd tailing integrates with trace IDs; Zipkin compatibility bridges legacy instrumentation.
Roadmap Acceleration: Plans for gRPC transport, adaptive sampling and service dependency graphs gained community momentum post-admission.

Tracing Best Practices

Enable baggage propagation carefully; large payloads impact latency.
Sample at ingress (Envoy) and propagate sampling decisions downstream to avoid partial traces.
Annotate spans with Kubernetes metadata (namespace, pod, node) to correlate with cluster events.
Pair Jaeger with Grafana/Prometheus dashboards to pivot between traces and metrics during incident response.

Summary

Aspect	Details
Announcement Date	September 13, 2017
Key Innovations	CNCF backing, OpenTracing alignment, Kubernetes-native deployment patterns
Significance	Established Jaeger as the community-standard tracing stack for cloud-native teams

Table of Contents