Shared Nothing Architecture - How to Scale to Infinity
summary
A scheduling app built in Ruby on Rails runs isolated per-event deployments on fly.io, with one-VM cells, separate SQLite/Postgres databases, Redis for websockets, and Fusion Passenger routing. Print jobs use Puppeteer on separate on-demand 2GB machines that spin up in ~200 ms. Backups use rsync to replicate idle DB files to all regions and to a home server; logs are centralized. It scales to 94 events in 42 cities in 5 countries with sub-100 ms local responses and resilient failure isolation.
AI-written summary · about
chapters (7)
- 0:00Introduction and background
- 7:54Shared nothing architecture basics
- 12:29Scaling with multiple machines
- 17:45Optimizing for user proximity
- 20:17Macro services and backups
- 24:46Administration and automation
- 26:30Q&A session
related talks
46mRuby in the Cloud
27mScaling a Rails Monolith w/Event Driven Microservices
46mScaling With Friends
33mKeynote: Rails Doesn't Scale
01h 03mThe Trail to Scale Without Fail: Rails?
27mScaling Ruby Applications: Challenges, Solutions and Best Practices
42mHow Rails fosters a diverse and competitive tech ecosystem in the era of big tech
34mIndustrial Rails