技术分享：Prometheus是怎么存储数据的（陈皓） скачать в хорошем качестве

技术分享：Prometheus是怎么存储数据的（陈皓）

Slides: https://www.slideshare.net/haoel/how-... 0:00 1. Opening Remarks 1:50 2. Self Introduction 2:00 3.1 Understanding Time Series 2:20 3.2 Understanding Time Series Data 5:28 3.3 2D Data Plane 8:00 3.4 The Fundamental Problem 12:08 4.1 Prometheus Solution (v1.x - "V2) 15:05 4.2 Series Churn 16:24 5.1 New Design of Prometheus (v2.x) 16:32 5.2 Fundamental Design - V3 20:43 5.3 Blocks - Little Database 21:48 5.4 Tree Concept 22:22 5.5 New Design's Benefits 22:39 5.6 Chunk-head 24:15 5.7 Chunk head - Block 25:20 5.8 Large file with "mmap" 26:58 5.9 Write-Ahead Log (WAL) 30:03 5.10 Prometheus WAL & Checkpoint 34:43 5.11 Block Compaction 35:54 5.12 Retention 37:24 5.13 V2 - Chunk Query 37:30 5.14 V3 - Block Query 37:57 5.15 V3 - Retention 38:14 5.16 New Design's Benefits 40:17 5.17 Index 42:12 5.18 Sets Operation 44:15 5.19 Sort the Array 46:18 6. Benchmark (v1.5.2 vs v2.0) 48:00 7.1 (Facebook Paper) Gorilla: A fast, scalable, in-memory time-series database 50:55 7.2 Fundamental (Delta Encoding) 52:27 7.3 Compress of timestamp - Delta-of-Delta 53:33 7.4 Compression Algorithm 57:57 7.5 Sample Compression 58:42 7.6 Open Source Implementation 59:34 8. Reference