Distributed Tracing in the Wild
summary
Shows concrete deployments: Zipkin + Spring Cloud Sleuth used for local dev while hotels.com proxies traces to Haystack with a Pitchfork proxy; a real Typeform trace is 50.5 ms across 3 services with depth 6 and 10 operations; Netflix converts legacy formats to Zipkin/B3 and runs a Flink pipeline into Elasticsearch/S3 with feature-flagged 100% sampling for experiments; Yelp and SoundCloud use hybrid sampling (0.01% vs 100% local firehose) and reactive apps show the highest tracing overhead (~30% in one reactive benchmark).
AI-written summary · about
chapters (27)
- 0:00Intro
- 0:53Who am I
- 1:47What is Distributed Tracing
- 3:49Why do people care
- 6:18What is tracing
- 10:35Spring Cloud Sleuth
- 12:52Single File Deployment
- 13:29What is a Site
- 14:42Questions
- 16:24Sitespecific tags
- 17:35Why bother
- 20:50Ideal Application Environment
- 22:04Is a Zipkin Server in Use
- 24:27Different Libraries
- 25:40Case Study
- 27:39Netflix
- 29:13Flink Pipeline
- 30:04Stellar
- 31:07Metadata
- 33:59Data Collection
- 35:10Service Catalog
- 36:16Secret Scrubbing
- 37:35Data Retention
- 38:46Service Naming
- 39:54Data Policy
- 40:23Archetypes
- 41:58Links
related talks
30mDistributed Tracing: From Theory to Practice
24mDistributed Tracing : New DevOps Foundation.
01h 11mDistributed Tracing : Latency Analysis for Your Microservices
33mDistributed Tracing - Implementing Latency Analysis for microservices on Cloud Foundry
34mDebugging Your Services with Distributed Tracing
42mAn introduction to distributed tracing and Zipkin
29mDistributed Request Tracing
46mPintrace: A Distributed Tracing Pipeline