Opscode is a new company built for a new breed of IT professional. The Seattle startup says its community of developers is creating a pool of open source tools for developer operations (DevOps), a combination of a systems administrator and developer who is required to quickly configure and update systems in dynamic data center environments.
Opscode is led by executives who have been responsible for managing some of the world’s largest and highly dynamic computing infrastructures. As Master of Disaster at Amazon.com, Opscode Co-Founder and CEO Jesse Robbins was responsible for Web site availability for every property bearing the Amazon brand, while Opscode VP of Engineering Christopher Brown was a founding member, architect and lead developer for Amazon’s EC2. Co-founder and CTO Adam Jacob also founded HJK Solutions, an automated infrastructure consultancy.
Beyond the ‘SysAdmin’
“The importance of the operating system and low-level configuration is being pushed away as people have standardized on platforms,” says Robbins. “Now the next-generation approach to infrastructure is more dynamic, as we get into virtualization and cloud computing. In a dynamic infrastructure, the SysAdmin is required to think in a different way when there are shorter release times.”
In a traditional infrastructure, it may take months to get a server online. SysAdmins in a dynamic infrastructure, such as those at Amazon.com, Yahoo, and Google, are required to have new servers up and ready in a matter of minutes. They’re also required to have some software development knowledge to be able to quickly configure software for the systems.
‘Chefs’ Cooking in the Cloud Kitchen
Opscode, which in April won $2.5 million in a Series A funding round led by Draper Fisher Jurvetson, has developed Chef, an open source framework that relies on a community of developers (Opscode calls them “Chefs”) to write configuration management software. Chefs write “Recipes” that describe how DevOps want a part of their servers, such as Apache, MySQL, or Hadoop, to be configured.
The Recipes describe a series of resources that should be in a particular state, explains Opscode on its web site.
In a typical user environment, the Recipes would sit on a dedicated server, which would distribute the Recipes to each node that the DevOps needs to configure. Each node runs a client copy of Chef to process the Recipes. The configuration of the nodes are automated.
Public “Cookbooks” are categories, such as databases, process management, and Web servers, to which Chefs have submitted their Recipes. Public Recipes are rated by other users, and a link is provided to the Recipe’s maintainer.
50 Developers Involved
Opscode says it is measuring its success by the rate of activity in its community. Since the community was launched a month ago, Opscode says it has 50 developers contributing, with a further 100 signed up. Its users come from Web companies, including Ruby on Rails creator 37signals, analytics company Webtrends, and cloud computing provider RightScale.
Robbins acknowledges that automated configuration management tools exist but he believes those are targeted at legacy SysAdmin and systems engineering job roles. “SysAdmins are not just responsible for patch-level updates,” he says. “They’ve moved up the stack. There’s now a tight integration between SysAdmins and developers and our tools enable that.”
Robbins says that dynamic infrastructure sites such as Amazon.com, Google and Yahoo have built proprietary management technologies to enable infrastructure automation. Robbins says Opscode was established to bring such automation tools to the masses.
Opscode is expected to officially launch in the market by the end of the first quarter, when it will also reveal its revenue model.