From Rails Rumble to 50,000,000 results
summary
StillAlive grows from a 48‑hour Rails Rumble prototype to a service reporting 50,000,000 results (65M by talk time) and explains the concrete fixes that make that scale: replace mechanize with capybara-webkit for JS checks while blacklisting ad domains, fix a four‑minute query by adding a compound concurrent INDEX on (site_id, created_at) to drop it to 0.448 ms, swap heavy delayed_job workers for lightweight distributed workers using ZeroMQ queuing, automate with Chef and zero‑downtime Unicorn deploys.
AI-written summary · about
chapters (8)
- 0:00Introduction and background
- 3:20What is Still Alive
- 5:08Building features and scaling
- 10:36Feature scalability
- 18:49Performance monitoring
- 23:49Database optimization
- 31:06Queuing systems with ZeroMQ
- 42:53Process scaling and summary
related talks
37mThe Recipe for the Worlds Largest Rails Monolith
01h 03mRuby at Scale
48mOn the Edge of Rails Performance
14mDatabase Performance at Scale for RoR Applications
33mServices and Rails: The Sh*t They Don't Tell You
35m"10 performance sins of Rails applications"
37mScaling Rails for Black Friday and Cyber Monday
26mMonolith-ifying perfectly good microservices