Improving the efficiency of more than 800 databases with o11y
summary
OVHcloud builds an internal DB service that wires Postgres and MySQL logs and metrics into OpenSearch + Mimir + Grafana, parses slow queries by threshold, and enforces schema checks in CI to catch missing or excess indexes. Hardware is vertically scaled and synthetic probes use patched sql runners scrapped by Prometheus histograms. Outcome: from a peak of 2 million slow queries on one DB to under 1 million across 1,200+ databases, with SLO-based health aggregation per cluster.
AI-written summary · about
chapters (10)
- 0:00Introduction and Speaker Background
- 1:13Overview of OVHcloud's Database Infrastructure
- 2:36Challenges in Database Performance and Observability
- 3:44Implementing Observability Solutions
- 5:34Optimizing Database Performance
- 6:49Addressing Slow Queries and Schema Optimization
- 11:29Adopting SLO Framework for Databases
- 14:04Synthetic Monitoring and Quality of Service
- 17:37Future Plans and Conclusion
- 18:49Q&A Session
related talks
46mBig databases, many servers, data center or cloud - it's all here!
24mDevOps. Monitoring Databases. On-Premise and Cloud environments
32mManaging Databases in the Cloud in Broken
31mUsing Databases to pull your applications weight
48mMonitoring OVH 300k Servers, 27 DCs: A Metrics Platform
41mMonitoring OVH: 300k Servers, 27 DCs, a Metrics Platform
47mThe Road Towards an Open Database Performance Ranking
43mDatabase in a SaaS application