Cognitive and Self-Adaptive System for Effective Distributed-Tracing...
summary
Applications keep existing Jaeger/Zipkin instrumentation while libraries send spans to a span aggregator that merges partial spans, clusters traces by density-based spatial clustering, and applies an adaptive tail sampler. Sampling before is typically under 5% but the system increases capture of error/unusual traces to about 35–40% and reduces storage needs by 25–27% in large clusters. Implementation is production‑deployed, tunable, and layered on top of standard Dapper/Zipkin/Jaeger pipelines.
AI-written summary · about
chapters (7)
- 0:00Distributed tracing challenges
- 1:10Current tracing implementations
- 2:35Random sampling problems
- 4:44Tail-based adaptive sampling
- 6:07Adaptive sampling logic
- 9:16Production results and benefits
- 10:44Future improvements
related talks
14mCognitive and SelfAdaptive System for Distributed Tracing
13mCognitive and Self-Adaptive System for Effective Distributed-Tracing- 14mLightning Talk: Cognitive and SelfAdaptive System for Distributed Tracing
13mDistributed Tracing: Adaptive and Telemetry-Based Approach for Effective...
49mDistributed Tracing: How the Pros Debug Concurrent and Distributed Systems
57mDistributed Tracing: How the Pros Debug Concurrent and Distributed Systems
41mTracing Real-Time Distributed Systems
59mDistributed Tracing and Monitoring with OpenCensus