Jaeger Deep Dive
Talkat KubeCon + CloudNativeCon North America 2019 · on Wednesday, 2019-11-20· at 14:25· Room 16AB - San Diego Convention Center, Mezzanine Level·
summary
Explains Jaeger internals: Go collectors and UDP/agent or sidecar forwarding, storage choices (Cassandra, Elasticsearch, Badger local), React UI, and multi-language clients; adds a Kafka-backed pipeline with an ingester/indexer to absorb spikes and enable Flink streaming instead of expensive Spark batch jobs; describes sampling strategies—probabilistic, rate limiting, endpoint-level, experimental tag-matching, and planned tail-based sampling (Uber rollout in progress); shows Jaeger Operator/Helm, a gRPC storage plugin model, deep dependency graphs, and a trace DSL plus Jupyter workflow for analytics.
AI-written summary · about
chapters (15)
- 0:00Intro
- 1:27Jaeger, a Distributed Tracing Platform
- 2:45Technology Stack
- 3:37Project & Community
- 9:47Joeger vs. OpenTracing
- 16:05Sampling
- 21:20Integrations with other CNCF projects
- 23:30Architecture 2017: Push
- 25:07Asynchronous span ingestion
- 25:56Architecture now: Push+Async+Streaming
- 28:32Zipkin Compatibility
- 29:26New Features
- 30:42External storage plugins
- 31:24Graph Visualization of Multiple Traces
- 34:09Roadmap
