By Adrian Otto
Distinguished Architect, Rackspace, and OpenStack Magnum PTL
Rackspace is proud to support the 2015 OpenStack Silicon Valley event as a headline sponsor. Since co-founding OpenStack with NASA in 2010, Rackspace has been operating OpenStack in our public cloud, which has grown to one of the world’s largest. Operating any cloud at a global scale is a significant operational achievement. Come to our “OpenStack Fleet Management: What Rackspace Has Learned” session for unique insight from Matt Van Winkle, Senior Manager of Operations at Rackspace, into our experience over the past five years of running our OpenStack public cloud at scale.
Matt will share how we overcame scaling challenges in OpenStack as we evolved the software with a community comprised of contributors from 140 different affiliations. Deploying and tracking huge numbers of nodes and managing how they interact without a clear methodology is complex. We will share tips about how to monitor and deal with these challenges before they become problems. We will also explain how to balance the needs of offering both virtual machine and bare metal flavors. We will explain how we use containers in our fleet management as well.
Trusting Software to Perform Mission-Critical Workloads
Any significantly complex software system takes about five years to develop and mature to the point where there is significant adoption, and where edge error cases are fully addressed to the extent that the system is ready for mission-critical workloads. This rule of thumb applies to database software, cloud software, global monitoring software, and so on.
Would you pilot an unmanned spacecraft using software you dreamed up and wrote in a couple of months? What if a small mistake could cause you to lose a billion dollars worth of work, spiraling out into space with no way to ever get it back? That’s why we don’t do that. The truth is that complex systems take time to perfect, but it’s also true that there are workloads that are appropriate to run on more experimental software. We think you should have a variety of choices to fit the right workloads to the right technology to balance costs and risks.
OpenStack: Five Years In
OpenStack just celebrated its fifth birthday this summer. It has evolved considerably over this time, and we are very proud of what it has become with the help of such a vibrant community. We are doing our part to lead within the OpenStack community to add the transformative technology we will use in years to come.
During the Thursday keynote, I will participate in the panel discussion titled “Containers: A Rapid-Fire Reality Check” where we debate the respective merits of virtual machines and containers, and detail how they are different. I will provide an update on the OpenStack Magnum project, and will join the discussion about enabling wider adoption of containers.
Magnum, Containers, and Choice
Magnum offers OpenStack cloud operators and OpenStack users a choice of what container orchestration software to use. Whether your users want to use Kubernetes, or Docker Swarm, or even Apache Mesos, there are now ways to make any or all of them work together as first class resources in OpenStack.
You can use Magnum to make containers available as a service in your own OpenStack cloud. With container management software changing so quickly, there is risk in choosing just one as the central component of your container strategy. Early leaders are not always long term winners. Magnum gives you a way to manage this risk, and plug in the choices that make the most sense for your cloud users. When a long term leader emerges, you will have confidence that you can offer that prevailing choice without re-tooling your cloud.
We are honored to be a part of the OpenStack community, and look forward to meeting you in Mountain View. We’d like to hear about your interest in OpenStack, and what challenges you face in delivering your cloud strategy. What is your container strategy, and what do you plan to use containers for? What challenges do you face running your cloud at scale? Please visit our exhibit, find our top experts, and talk with us about it.
Photo by Steve Jurvetson.