Monitoring Microservices The Right Way
summary
Argues that Graphite+StatsD breaks under microservices and promotes labeling model with Prometheus-style labels, shows metric endpoints at /metrics and metric scraping and auto-discovery in Kubernetes, demonstrates scaling strategies by remote_write to long‑term stores and federating instances, names Thanos and Cortex as workable backends and Logz.io as a managed option, and quantifies the problem as massive high‑cardinality time series rather than vague “lots of metrics.”
AI-written summary · about
chapters (10)
- 0:00Introduction
- 1:45StatsD to modern needs
- 3:11Microservices explosion
- 4:56Kubernetes monitoring
- 5:59Third-party platforms
- 7:03Labeling models
- 9:55Metric scraping challenges
- 12:45Scalability for time-series
- 14:09Monitoring microservices challenges
- 15:54Q&A
related talks
24mMonitoring Microservices The Right Way
35mMonitoring Microservices
28mMonitoring Microservices w. Prometheus & Tracing Apps w. OpenTracing
42mMicroservices are the answer - but what is the question?
37mKnee Deep in Microservices
27mObservability with Prometheus and beyond
10mMonitoring Microservices: A Challenge
48mMetrics All the Way: Data Driven Operations