Introduction to Distributed Tracing
summary
An introduction to distributed tracing that claims a cheap, scalable storage model by flushing trace blocks to object storage and a Parquet-based backend, enabling long retention and columnar queries; shows a real cluster ingesting one petabyte a month with replication factor three; describes a new query language TraceQL focused on spans and tree structure and a reworked backend called Tempo 2.0; demo runs via docker‑compose with Grafana, Tempo, Prometheus, fake services, service graph, Trace→metrics generation and trace-to-log linking.
AI-written summary · about
chapters (25)
- 0:00Distributed tracing intro
- 1:28What tracing is and why it matters
- 5:23Steps to implement tracing
- 8:15Grafana Tempo intro
- 8:36Tempo storage efficiency
- 9:20Minimal Tempo setup
- 11:04Scaling Tempo to petabytes
- 11:46Why TraceQL was created
- 12:51TraceQL advantages
- 14:12Operating spans with TraceQL
- 15:38Aggregates in TraceQL queries
- 17:23Advanced TraceQL features
- 18:28TraceQL structural operators
- 19:08Tempo storage with Parquet
- 20:32Row vs columnar vs hybrid storage
- 22:39Correlating traces with logs
- 23:00Service graphs and trace analysis
- 23:21Generating metrics from traces
- 24:45Live demo setup
- 25:50Exploring trace metrics
- 26:34Querying spans in Grafana
- 27:18From metrics to traces
- 28:03Analyzing traces with node graphs
- 28:44Finding errors with service graphs
- 29:06Closing remarks
related talks
45mMastering Distributed Tracing: Insights from Building a Platform from Scratch
24mDistributed Tracing : New DevOps Foundation.
16mDistributed Tracing is Awesome, Testing is Not... Until Now
49mDistributed Tracing: How the Pros Debug Concurrent and Distributed Systems
37mObservability, Distributed Tracing & the Complex World
34mReduce Complexity with Distributing Tracing
59mAn introduction to distributed tracing
22mDistributed Tracing