Jaeger Distributed Tracing Platform

Jaeger (German for hunter, pronounced yay-gr) is an open-source distributed tracing platform for monitoring and troubleshooting distributed systems and microservices architectures. Originally developed at Uber Engineering (by Yuri Shkuro and the Observability team, headquartered in San Francisco) starting in 2015 to address the challenges of debugging request flows across hundreds of microservices, Jaeger was accepted into the Cloud Native Computing Foundation (CNCF) in 2017 and graduated in October 2019, becoming the fourth CNCF graduated project after Kubernetes, Prometheus, and Envoy. Key features: distributed tracing: tracks requests as they flow through distributed systems, recording timing data and metadata at each service hop, creating end-to-end traces that show the complete path and timing of a request. Trace structure: each trace consists of spans (units of work), organized in a parent-child tree representing service calls and internal operations, with each span containing operation name, start time, duration, tags, logs, and baggage. Context propagation: propagates trace context across service boundaries using headers (W3C Trace Context, Jaeger native, B3/Zipkin, Uber headers) over HTTP and gRPC. OpenTelemetry support: native integration with OpenTelemetry SDK and OpenTracing API. Storage backends: pluggable storage with support for Elasticsearch, Cassandra, Kafka (ingestion pipeline), BadgerDB (embedded, for testing), and in-memory storage. Adaptive sampling: dynamic sampling that adjusts sampling rates based on traffic patterns. UI: web interface for searching, viewing, and analyzing traces, with a flame graph visualization, dependency graph showing service relationships, and deep-linking to logs. Metrics: Prometheus integration for collector and query metrics. Alerting via Prometheus AlertManager. Cross-platform: Docker, Kubernetes (via Jaeger Operator). Apache-2.0.