Cache = Cash! 2.0
summary
Claims concrete wins: focusing on CPU caches (L1/L2) and data layout, switching heavy ActiveRecord loads to trimmed SELECTs and smaller SQL types, adding missing indexes, fragment caching with touch updates, and HTTP ETag/Last-Modified checks. Benchmarks on a Raspberry Pi Model 3 with 50,000 seeded products (generated by Claude/agent) show micro‑optimizations +3% and +6% then a 60% win from proper touch keys and static prerendering (50,000 static product files), taking the workload from ~30,000 ms down to ~3,500 ms — a roughly tenfold throughput gain.
AI-written summary · about
chapters (10)
- 0:00Introduction
- 2:09CPU Cache
- 6:08Back to 2013
- 7:18Back to 2025
- 9:44Baseline
- 15:42Fragments
- 17:24HTTP Cache
- 22:53Autoban
- 26:42Speed
- 28:07Webpage comparison
