Intro: Cortex
summary
Cortex provides horizontally scalable, highly available, multi-tenant long‑term storage for Prometheus metrics, ingesting via the remote_write API and persisting to backends like DynamoDB/BigTable backends or Cassandra (BoltDB for laptop). It replicates incoming samples (typical replication factor 3) so one or two nodes can fail without gaps, and runs many Prometheus instances — examples cited of 30–40 Prometheuses feeding one cluster — while the query layer shards work across processes; the demo boots local Docker nodes, joins a two‑node ring, and works.
AI-written summary · about
chapters (13)
- 0:00Intro
- 0:30Show of hands
- 1:45Questions
- 2:38Cortex
- 3:58People using Cortex
- 6:40Global View
- 9:10Horizontal Scalability
- 12:17Replication
- 13:43Storage
- 17:27BigTable
- 21:00Multitenant
- 25:29Tom Wilkie
- 28:59Airplane Mode
related talks
28mCortex 101: Horizontally Scalable Long Term Storage for Prometheus
20mIntro to Scaling Prometheus with Cortex
45mScaling Prometheus with Cortex, one time series at a time
25mCortex: Ingesting and Querying 100s of Billions of Samples Per Day
16mCortex: Multi-tenant Scalable Prometheus
20mCortex: Intro and Production Tips
41mWeave Cortex: Horizontally Scalable, Multitenant Prometheus as a Service
22mCortex: Ingesting and Querying 100s of Billions of Samples Per Day