Percona Community Live Conference - Day 1 - Open Source Databases Conference
summary
Claims that VictoriaMetrics stores huge time series cheaply and with far lower CPU than Prometheus, demonstrated by a 12‑hour test that ingests 344 billion metrics and uses ~9.67 GB storage vs Prometheus ~10.49 GB while CPU drops from ~74% to ~32%. Shows actual tools and ops: vmagent, push mode, deduplication and retention policies, and explains configuring downsampling and duplication windows. Result: a working, repeatable testbench and concrete knobs for long‑term metrics retention.
AI-written summary · about
chapters (13)
- 0:00Percona Community Live intro
- 5:22Database resiliency
- 1:05:09Percona distributions
- 2:00:58MongoDB data wrangling
- 3:04:53Emulating block device incidents
- 4:04:53TiDB overview
- 5:04:44VictoriaMetrics storage
- 6:04:44Open source DB trends
- 7:04:50Percona Server on Pi
- 8:05:06Point-in-time recovery with pgBackRest
- 9:05:39Database design
- 10:06:03Liquibase for version control
- 11:06:07ProxySQL 2.4 features
related talks
40mScale Your Monitoring Solution With the VictoriaMetrics Ecosystem
36mVictoriaMetrics: scaling to 100 million metrics per second (EN)
34mSpecifics of Data Analysis in Time Series Databases
50mHow To Improve PostgreSQL Monitoring and Observability With PMM
30mHow We Scaled Metrics Monitoring for 10,000 Applications With Victoria...
30mWhy we recommend PMM to our clients
46mPrometheus as a Time Series Database
53mExtending Percona Monitoring and Management With Custom Collectors