How Shopify Scales Rails
summary
Shopify claims one Rails codebase evolves to handle 40k stores and spikes to ~500k sales/day by adding layers not rewrites: cachable page caching (gzipped, generational keys), identity cache for model read‑through memcached objects, backgrounding via Resque to remove work from Unicorn processes, and a custom load tester Conan. Monitoring stack is New Relic, Splunk, statsd; DB work uses Percona tools, MySQL tuning and TCMalloc. Measured wins: 10–15x DB speed from in‑memory working set and big RPM reductions after caching.
AI-written summary · about
chapters (21)
- 0:00Intro
- 0:23Overview
- 2:12The Stack
- 6:00Looking Back, to Look Ahead
- 10:20Know The System
- 12:05Chive Flash Sale
- 13:03Measure ALL THE THINGS
- 14:26New Relic
- 14:53Splunk
- 16:52cacheable
- 19:09Caching Dynamic 404s
- 20:22Identity Cache
- 23:34Delayed Job
- 24:17Resque
- 26:47Redis
- 28:15MySQL Hardware
- 28:51MySQL Query Optimization
- 30:24MySQL Tuning
- 33:51after_commit
- 36:10Services
- 37:06Imagery
