Joyent Scale Services
Joyent’s developers and system architects have previous experience in many large-scale email, Web, newsgroup, database and computational clusters, and have deployed and scaled multi-component applications written in C, C++, Java, Python, Ruby, Erlang, Perl and more.
In addition, Joyent has developed, built and scaled some of the earliest Ruby on Rails applications – and as a result, developed a world-class infrastructure, a methodology around how to deploy and scale (both up and down) Rails applications and most importantly, a set of infrastructure geared specifically towards assessment, benchmarking, and introspection. The benchmarking infrastructure is stand-alone and validated, and introspection makes extensive use of DTrace.
You can leverage this combination of insights into development, application design and system architecture to ensure you don’t make many of the early mistakes. This is key to making sure that your development workflow isn’t affected by system issues, and that your end-users always have a great experience.
Joyent’s Scale services provide you with world-class consulting to get your application up and running at peak efficiency. The scale group reports to the CTO, Jason Hoffman, a noted expert and frequent conference speaker (RailsConf Europe, 2006, RailsConf 2007, RailsConf Europe 2007).
The group does per diem engagements and has a developing line of packaged consulting solutions. The current packaged consulting solution is a complete top-to-bottom audit of Ruby on Rails applications.
Audit of a Ruby on Rails application stack
The goal of the Ruby on Rails application audit is to align your functional requirements for your application with a predictably scalable systems architecture. A critical part of our audits is the extensive use of our DTrace providers at every level (serving, application, database), and our ability to identify and patch code in the operating system, web, application and database tiers that are negatively impacting one’s application performance (for example, we have found bugs in Solaris, Rails, Ruby and MySQL, the fixing of which solved the problems).
Representative results and successes from a single application:
- Identification of a file directory structure that wouldn’t scale for any purpose, and the suggestion of converting to a 256×256 hashed directory structure with LDAP.
- Reduction of CPU use by 80% while increasing the load 5x by identifying several CPU-intensive hotspots.
- Rapid identification of a memory leak.
- Identification of custom functionality as the bottleneck. This was then moved into a Ruby-C extension.
If you would like someone from Joyent to contact you regarding your requirements please submit the following:
