Observability: Dust Off Prometheus with VictoriaMetrics
summary
VictoriaMetrics is presented as a drop-in replacement for a Prometheus+Thanos observability stack to reduce resource use and cost: benchmarks show up to 7x less RAM versus Prometheus+Thanos/Cortex (10x vs InfluxDB) and an actual migration reduced Prometheus+Thanos footprint by ~4x. The architecture separates lightweight agents and ingestion/query services (VM Agent -> VMInsert -> VMSelect), runs as a Kubernetes cluster with configurable long-term cold storage (default 24h, extendable to years), and is validated in production scraping ESP8266 temperature probes.
AI-written summary · about
chapters (15)
- 0:00Introduction
- 1:35Monitoring to observability
- 3:00Metrics and time series
- 5:27Prometheus issues
- 7:56Thanos vs Cortex
- 8:59VictoriaMetrics overview
- 11:25Performance benchmarks
- 12:50Simple vs cluster architecture
- 14:14Data collection and querying
- 16:44Alerts and migration
- 17:46Security and UI
- 19:12Downsampling and Kubernetes
- 20:17Resto du Cœur case study
- 23:25Why VictoriaMetrics
- 25:16Thanks and Q&A
related talks
40mScale Your Monitoring Solution With the VictoriaMetrics Ecosystem
27mObservability with Prometheus and beyond
36mObservability with Prometheus and Beyond: Leveraging Cloud-native Technologies Outside of Cloud
28mHow to reduce expenses on monitoring with VictoriaMetrics
37mThanos: Prometheus in HA Mode with Persistent Storage
49mObservability Day | Project Updates
36mVictoriaMetrics: scaling to 100 million metrics per second (EN)
26mPrometheus That Scales