OLTP or OLAP: why not both?
summary
Covers Vitess history and architecture with concrete demos: a live Vitess resharding and demo, walkthrough of VReplication explained and a VReplication demo, plus numbers — hundreds of shards in production (256 at YouTube), tens of thousands of MySQL instances (about 25–35k) — and how features actually work: SQL parsing and rewrite, connection pooling, hot‑row protection, read‑after‑write vs eventual reads, planned and unplanned reparenting, and the exact resharing workflow.
AI-written summary · about
chapters (16)
- 0:00Introduction
- 1:09About me
- 2:30What is WIT
- 4:20Features and capabilities
- 6:48Built for cloud
- 8:57Vitesse stats
- 9:12Who is using Vitesse
- 10:46The problem Vitesse solves
- 12:01Vitesse architecture
- 19:43Replicas
- 28:15Master Promotion
- 32:02Vertical Rerouting
- 33:42How does it work
- 35:21Recharging
- 36:20Failure Domains
- 40:06Demo view
related talks
28mVitess: Introduction and New Features
40mVitess: Introduction And New Features
34mInside Vitess: Building Resilient, Distributed Databases
42mVitess VReplication: Standing on the Shoulders of a MySQL Giant
51mVitess beyond YouTube
44mIntro: Vitess
06mProject Lightning Talk: Vitess in Production: Powering Applications at Massive Sca...
45mDeep Dive: Vitess
