Cloud-native Metric Monitoring with Prometheus
summary
Prometheus is presented as the cloud-native, metrics-first monitoring stack with pull-based model and built-in service discovery, supporting 64-bit time-series, histograms and exemplars for linking metrics to traces. Claims: a laptop can ingest >1M samples/sec, ~200–250k samples/sec per core, largest Prometheus at 125M active series. Grafana ecosystem adds Mimir scaling to 1B active series (≈1,500 machines, 7,000 CPUs, 30 TB RAM), Grafana clusters ingest 165 TB/day and 2.2–5M samples/sec with 99% queries <2.5s. Loki indexes labels only, turning 10 TB logs into ~200 MB index.
AI-written summary · about
chapters (19)
- 0:00Intro
- 2:14Buzzword alert!
- 5:30Complexity
- 6:33Services
- 7:33SRE, an instantiation of DevOps
- 9:07Shared understanding
- 9:42Alerting
- 11:22Prometheus 101
- 12:36Main selling points
- 13:50Concepts & guarantuees
- 15:07Time series
- 18:37New features
- 20:10Cloud native defaults
- 20:53Prometheus scale
- 23:21Loki 101
- 25:17Loki @ Grafana Labs
- 29:56Tempo @ Grafana Labs (2022-09)
- 32:55Logs to metrics, the savings
- 35:26From logs to traces
related talks
36mObservability with Prometheus and Beyond: Leveraging Cloud-native Technologies Outside of Cloud
39mHow To Take Prometheus Planet Scale: Massively Large Scale Metrics Installations
27mObservability with Prometheus and beyond
10mPrometheus for Monitoring Metrics - Database Monitoring Tools
30mIntro: Prometheus
32mDemocratizing Application Monitoring with Prometheus
33mPrometheus Project Intro
42mPrometheus Intro and Deep Dive