Oh S***: How to bring a big Rails website down (and how not to)

on YouTube ↗34m

" summary

Scribd-scale numbers (≈40M monthly users, 50k uploads/day, tens of millions of rows) expose concrete Rails+MySQL failures: a quick change to validates_uniqueness plus lower(login) turns signup validation into a 2‑minute query; find_in_batches breaks on composite primary keys (fixed by a find_in_batches monkey patch); association.delete_all can null FKs instead of removing rows so use ActiveRecord delete_all when you need a SQL DELETE; missing composite indexes (eg. on user_id,deleted_at) forces use of USE INDEX/ FORCE INDEX or a functional index in Postgres; always run EXPLAIN and test on large datasets.

AI-written summary · about

" chapters (7)

1. 0:00Start
2. 1:15A bit about Scribd
3. 2:47Rules of thumb
4. 6:02The problem with find_in_batches
5. 10:11The problem with validates_uniqueness_of
6. 19:33The problem with delete and destroy
7. 25:46The problem with indexes

" related talks

• 26mBiggish Data With Rails and PostgresqlStarr Horne · Cascadia Ruby 2014
• 20mCRUD! The Consequences of Not Understanding How ActiveRecord ...Eileen M Uchitelle · MountainWest RubyConf 2014
• 29mScaling Rails to 2M MySQL requests per secondEugene Kenny · San Francisco Ruby Conference 2025
• 34mToo Big to FailChris Maddox · RailsConf 2014
• 43mSecuring Your Rails AppJim Weirich, Matt Yoho · Ruby on Ales 2011
• 48mOn the Edge of Rails PerformanceGregg Pollack · RubyConf 2009
• 31mA RAILS PERFORMANCE GUIDEBOOKCristian Planas · Euruko 2023
• 37mCache is King: Get the Most Bang for Your Buck From RubyMolly Struve · Ruby Hack 2019