Observability Wiki
Open source

Jaeger

A CNCF-graduated, open-source distributed tracing system originally built at Uber, now rebuilt on the OpenTelemetry Collector core.

Last updated

What it is

Jaeger is an open-source distributed tracing system originally developed at Uber and open-sourced in 2016 to help debug latency and failure issues across Uber’s microservices. It was donated to the Cloud Native Computing Foundation (CNCF) in 2017 and reached Graduated status — the CNCF’s highest maturity tier — in October 2019, putting it alongside projects like Kubernetes and Prometheus.

In November 2024, the project shipped Jaeger v2, a substantial architectural rewrite that rebuilds Jaeger as a distribution of the OpenTelemetry Collector, with Jaeger’s storage backends, query APIs, and UI layered on top. This means Jaeger v2 natively speaks OTLP end-to-end, ships as a single binary instead of separate collector/query/agent processes, and inherits the OpenTelemetry Collector’s broader receiver/processor ecosystem. Jaeger v1 reached end-of-life at the end of 2025, so v2’s OpenTelemetry-native architecture is now the project’s baseline going forward.

Why teams choose it

  • Mature, vendor-neutral, and free. As a CNCF-graduated project with a decade of production use (starting at Uber’s scale), Jaeger has none of the lock-in concerns of a commercial tracing vendor and a stable governance model.
  • Now genuinely OpenTelemetry-native. The v2 rewrite removes the translation overhead that existed when Jaeger spoke its own wire formats alongside OTLP, simplifying pipelines for teams already standardized on OpenTelemetry instrumentation.
  • Flexible storage backend choice. Jaeger supports Elasticsearch, OpenSearch, Cassandra, ClickHouse, and an embedded Badger store for single-node setups, letting teams match the backend to their existing operational expertise.
  • Operational burden is real. Unlike managed tracing SaaS, you own the deployment, scaling, and storage backend yourself — a legitimate cost that needs weighing against tools like Grafana Tempo or managed vendors that abstract storage entirely.

Pricing model

Jaeger is free, open-source software (Apache 2.0 license) with no vendor attached and no paid tier — all costs are the infrastructure and storage backend you choose to run it against (e.g., an Elasticsearch or Cassandra cluster), plus your own operational time. Several observability vendors offer managed Jaeger-compatible ingestion or bundle Jaeger as part of a broader commercial platform, but the core project itself carries no license cost.

Alternatives

The most direct open-source alternative is Grafana Tempo, which takes a different architectural bet — object storage only, no dedicated index — and tends to be cheaper to operate at very high trace volumes. Zipkin is the other long-standing open-source tracer, generally considered simpler but less actively evolving than Jaeger. On the commercial/managed side, Honeycomb and Datadog offer tracing as part of broader observability platforms with far less operational overhead, at the cost of a recurring bill.

Visit official site →