Prometheus: The Unsung Heroes
summary
Local.ch serves about 4 million unique visitors monthly and a closed‑source Brocade load balancer runs as a cluster of four boxes whose internal metrics aren’t exposed. The engineering effort replaces a brittle syslog→carbon VM with lightweight log parsing using mtail (regex programs, timestamp decorators and counters for 200/302/500, brew‑time timers) and SNMP scraping via snmp_exporter plus its generator to map OIDs to pool and virtual‑server metrics, producing usable Prometheus graphs and alerts.
AI-written summary · about
chapters (16)
- 0:00Introduction
- 1:17Examples
- 2:10The Problem
- 3:14Load Balancer
- 5:15Anything
- 9:55Time Parse
- 11:00Python decorators
- 14:36Runout of emptyin
- 16:30SNMP exporter
- 18:03Object identifiers
- 18:26MIB
- 19:02Generator
- 19:40Generator File
- 20:41Demo
- 25:58Alternatives
- 27:04Outro
related talks
29m“If you Don’t Monitor your Infrastructure, you Don’t Own it!” Regain Control Thanks to Prometheus
30mPrometheus Is Good for Your Small Startup
33mIntro: Prometheus
10mPrometheus for Monitoring Metrics - Database Monitoring Tools
33mPrometheus Project Intro
40mThe Prometheus monitoring system
32mAdapting Brownfield Deployments to Cloud-Native Technology
42mPrometheus Intro and Deep Dive