Troy Toman, the Vice President of Cloud Infrastructure Products at Rackspace, provides an inside look at how Rackspace is thinking about continuous delivery.


Rackspace is a hosting company that provides hosting services to their customers on dedicated gear, private clouds and public clouds. They focus on their service, and the ability to offer that service in a consumable way for their customers.

Implementing continuous delivery on an infrastructure product has been a learning experience for the company over the last couple of years. A lot of established best practices focus on continuous delivery with multiple deployments a day — but Toman is running infrastructure on a very large scale (hundreds of thousands of virtual servers) and he had to transform those concepts significantly.

Currently, they pull from Trunk once a day, and try to deploy every one to two weeks. Learning from that process, they’ve been working with the community to adapt their processes to move toward monthly milestones, to think about doing automated tests from Trunk on a regular basis, and to basically work cooperatively to make continuous deployment an easier process not only for Rackspace, but for others in the OpenStack community.

They are still working to quantify the benefits of continuous delivery, but they are certainly able to move things into production in a faster way.

Toman says it allows Rackspace to stay on the leading edge, while staying close within the community. They’re exploring new territory, which is an amazing part about being in an open source project. It’s not the perspective of Rackspace alone. Rather, it’s the collective perspective of Dell, HP, Cloudscaling, Piston, Nebulus, etc. They all come together to discuss different points of view that they wouldn’t discover on any one company’s own turf.

Toman is excited about the pace at which they’re progressed over the last several releases (and we are too!).


Image credit: "Rackspace" by Scott Beale / Laughing Squid