Speed Science

on YouTube ↗41m

" summary

Shows concrete claims and fixes: reproduces slowness, uses a scientific loop of hypothesis + measurement, and ships tooling like derail benchmarks, allocation_tracer and memory_profiler to profile boot and requests. Diagnoses a disk IO bottleneck (raising Sidekiq concurrency from 30 to 4 speeds work) and a heavy mail gem parser that inflates boot RSS; lazy-loading or swapping parser removes swap, cuts RAM and makes the app “scream” fast.

AI-written summary · about

" chapters (12)

1. 0:00Introduction to speed
2. 1:36How to make things fast
3. 2:30Personal introduction
4. 4:35Case study: debugging slowness
5. 7:28Disk I/O bottleneck
6. 10:33Scientific method in software
7. 12:54Understanding Ruby memory
8. 18:59Memory usage in Rails
9. 21:58Optimizing gem performance
10. 27:40Detecting bad gems
11. 34:03Results and takeaways
12. 36:22Funday Friday proposal

" related talks

• 04mMaking things go fast inRobin Appelman · Nextcloud Conference 2016
• 01h 15mSpeedupRails, Speedup Your CodeAaron Patterson · Red Dot Ruby Conference 2014
• 01h 02mPerformanceJeremy Kemper · RailsConf Europe 2008
• 41mProfiling to make your Rails app fasterGannon McGibbon · RailsConf 2021
• 29mHow to PerformanceEileen M Uchitelle · Full Stack Fest 2015
• 33mI Made a Slow Thing FastBen Hamill · LoneStarRuby Conf 2013
• 39mDisk is fast, memory is slow. Forget all you think you knowDaniel Magliola · RubyConf 2019
• 35m"10 performance sins of Rails applications"Aleksander Dąbrowski · RubyC 2019