Automatically Making Dashboards Load 100X Faster
Talkat KubeCon + CloudNativeCon North America 2020 Virtual · on Friday, 2020-11-20· at 17:05· Intrado Virtual Event Platform,·
summary
Claims automatic dashboard speedups up to 100x by detecting slow Prometheus queries from query logs and generating pre-aggregations using recording rules or an M3 aggregation tier with roll-up and mapping rules. Example: a CPU metric with 16,000 series and 20s query time becomes ~230 series and 0.5s after aggregation. A provided high cardinality analyzer finds repeated expensive queries, estimates cardinality, and emits proposed recording or roll-up/mapping rules; recording rules support full PromQL but are expensive, roll-ups drop raw series at ingest for automatic speedup.
AI-written summary · about
chapters (6)
- 0:00Introduction and agenda
- 1:22Understanding the problem
- 5:20Using recording rules
- 8:16M3 aggregation tier
- 13:57Rule comparison
- 16:13High cardinality analyzer
related talks
23mAuto-suggesting and generating recording rules for Prometheus
11mAggregating, Alerting and Graphing on Millions of Prometheus T...
20mAggregating Metrics In-Flight: Challenges, Opportunities
37mPractical Capacity Planning using Prometheus
13mLightning Talk: A Drop-in System To Accelerate Metrics Observability by 100x Us...
35mDoorDash’s Journey From StatsD To Prometheus With 10 Million Metrics/Second
46mHow many dashboards does your Grafana have?
29mStreaming Recording Rules for Prometheus, Thanos, and Cortex Using...