Jaeger is a free, open source distributed tracing platform born at Uber and graduated by the CNCF. See features, architecture, pricing and alternatives.
Jaeger is an open source distributed tracing system that helps engineering teams understand how requests flow through microservices architectures. Originally built at Uber in 2015 to solve the company's own scaling pains, it was open sourced in 2017 and donated to the Cloud Native Computing Foundation, where it graduated to top-tier project status in 2019 alongside Kubernetes and Prometheus.
The project has since evolved to align tightly with OpenTelemetry, the industry standard for instrumentation, meaning Jaeger now ingests OTLP trace data natively and can be deployed as an OpenTelemetry Collector-based backend. It remains entirely free and self-hosted, run on infrastructure the adopting team controls rather than a vendor's cloud.
Jaeger's UI lets engineers drill into a single trace as a waterfall or flame graph, showing exactly which service call added latency or threw an error, alongside a service dependency graph that maps how microservices call one another across the whole system.
Under the hood it supports pluggable storage backends (Cassandra, Elasticsearch, OpenSearch, Kafka, and now ClickHouse), adaptive sampling to control data volume at high traffic, and broad language coverage through OpenTelemetry SDKs for Go, Java, Python, Node.js, C++, and more.
Jaeger itself has no pricing: it is licensed under Apache 2.0 and free to run in any environment. Costs come entirely from the infrastructure a team provisions to run the collector, query service, and chosen storage backend, plus optional third-party commercial support.
For teams that don't want to operate Jaeger themselves, several vendors offer managed OpenTelemetry-compatible tracing backends (such as Grafana Cloud Traces) that are wire-compatible with Jaeger's data model, letting organizations start self-hosted and migrate to a managed option later without re-instrumenting their code.
Yes. Jaeger is fully open source under the Apache 2.0 license, with no license fees or usage caps; you only pay for the infrastructure you run it on.
Jaeger was created at Uber Technologies, led by engineer Yuri Shkuro, and was open sourced in 2017 before being donated to the Cloud Native Computing Foundation.
Yes. Jaeger v2 is built on the OpenTelemetry Collector framework and natively accepts trace data via the OpenTelemetry Protocol (OTLP).
Jaeger supports Cassandra, Elasticsearch, OpenSearch, Kafka as a buffering layer, and, since version 2.18.0, native ClickHouse.
No. Jaeger is self-hosted software; there is no official managed cloud version, though several vendors offer OTLP-compatible managed tracing backends.
It is typically paired with Prometheus for metrics and a log aggregator such as Loki or the ELK stack to build a complete observability stack for microservices.