Deploy, Scale and Sleep at Night with JRuby

Your website has just crashed, and you're losing money. The application is built on Rails, runs on MRI, and is served up with Mongrel and Apache. Having this kind of infrastructure means that you're managing more processes than you can count on two hands. It will take a while to get back online because there are so many components in your system; including resque, monit, cron, Ruby Daemons, and more. But the website has to get up and running if you are going to stay in business.

The problem I've just described is all too common. It has happened to everyone from small start-ups to large social networking sites that use Ruby to serve millions of requests. But the recent growth and increased adoption of the Java Virtual Machine (JVM) as a platform for Ruby applications means that this is no longer the only way of doing things.

In this talk, I'll discuss the new architecture and deployment strategies that JRuby enables, and why they're better for your Rails applications. I'll talk about what makes JRuby more scalable and show some benchmarks that prove it. I'll also discuss three deployment strategies that can be used with any Rack-based JRuby application, and provide an overview of how each one works. I'll focus on the technologies that make them possible, and when it's appropriate to use each strategy. They include:

• Warbler: a gem can be used to create a self-contained web application archive file.
• Trinidad: a light-weight web server for creating flexible, modular deployments that still feel friendly and familiar.
• TorqueBox: an all-in-one environment that includes built-in support for messaging, scheduling, daemons and clustering.

Choosing the right JRuby deployment strategy will not only provide a powerful and reliable platform -- it will help you get more sleep at night.