Parting the Clouds: Tracing Cloud Foundry for Improved Performance...
summary
Shows concrete tracing wins for Cloud Foundry: adding OpenTracing middleware and LightStep/Jaeger/Zipkin backends exposes full request traces so operators see where time goes — cf apps CLI requests ~200ms with cathy ~180ms, one run shows go-router blocking for 8.5s and producing a 502, double-login in v3 CLI duplicates work and removing it reduces ~100ms per command, and tracing surfaces which downstream calls inflate latency. Mentions the OpenTracing/OpenCensus merge and vendor-neutral headers to stitch traces across systems.
AI-written summary · about
chapters (5)
- 0:00Start
- 10:03Service Level Objectives
- 10:37Service Level Indicators
- 17:25What Actually Is Observability
- 29:01What's Next on the Open Source
related talks
05mLightning Talk: That Small Difference Made all the Difference
33mDistributed Tracing - Implementing Latency Analysis for microservices on Cloud Foundry
46mVisualising and tracing requests through a cluster: Intergrating OpenTracing into OpenStack Swift
16mPragmatic execution tracing
19mOops my app performance sucks... what if observability could be a game changer?
24mWhen Connections are Magic: Understanding Performance in Serverless
33mOpenTracing and Containers: Depth, Breadth, and the Future of Tracing
50mDistributed Tracing in the Wild