Ten Years of Rails Deployment
summary
Timeline of Rails deployment from 2004 to 2013 names concrete shifts: CGI → FastCGI → Mongrel (common setup with three mongrel processes proxied by HAProxy) → Passenger (Apache first) → Unicorn (master + workers with zero downtime worker handoff) → Puma (multiworker multithreaded). Calls out gotchas: assets precompile needs ruby racer and libv8 binaries, Ruby install choices (package vs rvm), Docker for isolated Ruby upgrades and rollbacks, and a 150‑server case where SSH deploys become slow.
AI-written summary · about
chapters (15)
- 0:00Rails deployment history
- 3:58From cPanel to version control
- 5:322004 Rails deployment options
- 9:25Memory usage in early apps
- 10:29Mongrel and standalone servers
- 12:55Passenger simplified deployments
- 15:48Unicorn and zero-downtime
- 18:19Puma’s threaded approach
- 21:54JRuby Java deployment
- 22:59Capistrano basics
- 26:12Ruby setup on server
- 28:00Managing dependencies
- 29:50Docker containers
- 32:41Advanced tools like Mia
- 33:47Ansible example
related talks
26mFrom FTP to Kamal: 20 Years of Deploying Rails
31mTen Years of Rails Tutorials
46mHeroku 2014: A Year in Review
36mContinuous Deployment with Rails
34mLooking Backward: Ten Years on Rails
26mDeploying Rails is Easier Than it Looks
35mChanging the wheels on the bus at 80 mph: Upgrading to Rails 3
31mOperating Rails: What About After You Deploy?