Shipping Metrics From the Edge
Talkat KubeCon + CloudNativeCon North America 2019 · on Wednesday, 2019-11-20· at 11:50· Room 11AB - San Diego Convention Center, Upper Level·
summary
Edge Prometheus runs on tiny devices (k3s on Raspberry Pi) and uses remote_write to replicate Prometheus WALs into a scalable ingestion layer: a Thanos receive hash ring that distributes time series, writes 2‑hour TSDB blocks to object storage and serves recent samples directly. Default TSDB block writes every 2 hours and 15s scrapes can be lost if federating only every few minutes; the demo shows a Pi remote writing successfully and week‑old metrics queryable from the central store.
AI-written summary · about
chapters (18)
- 0:00Introduction
- 0:55The Problem
- 2:10The Setup
- 3:55Reverse Proxy
- 4:37Central Prometheus
- 5:29Prometheus Query
- 6:23Final Sidecar
- 7:33Remote Write
- 9:18Inverted Prometheus
- 9:49Improved Prometheus
- 13:43Multiple Stateful Sets
- 14:59Demo
- 17:22Summary
- 18:22Pharos
- 19:02Push vs Pool
- 19:21Alerts
- 20:45Remote Writing
- 21:20Air Gap Metrics
related talks
36mObservability with Prometheus and Beyond: Leveraging Cloud-native Technologies Outside of Cloud
10mUpstream-First, High Scale Prometheus Ecosystem
28mIntegrating Long-Term Storage with Prometheus [A]
42mPrometheus Intro and Deep Dive
01h 19mIntro + Deep Dive: Prometheus
40mBuilding a Scalable Observability Platform for Edge Compute Nodes
36mPrometheus - Intro, Deep Dive, And Open Q+A
33mPrometheus Project Intro