OpenStack Summits follow a six-month cadence and the first of this year’s shows is being held this week in Boston with OpenStack Sydney Australia following in November. Attendance here in Boston is said to be approximately 5,000 and the keynotes are filled to capacity. One of the things people are watching for is whether a handful of prominent enterprise IT vendors are backing away from their previous commitments to OpenStack. The reason I often hear is summed-up in one word—complexity. Viewed from the standpoint of their enterprise customers they see too many projects to integrate, too many release cycles to manage, coupled with too few IT staff members that have the proficiency to make it work.
Prior to this week’s event, I spoke to a number of enterprise hybrid cloud users who had hands on experience with OpenStack within their organizations. Of those, only one is progressing forward. The others have either put their projects on hold or abandoned them altogether. And yes, complexity comes up as a reason. But its more than that.
As I spoke to these people, it became apparent that they started with the assumption that using OpenStack was equivalent to using open source instead of VMware vSphere to manage virtualized IT environments. VMware was getting too expensive, they were feeling locked-in, and private clouds were on the horizon. So rather than spend even more to transform vSphere into a private cloud, the alternative was to spin up OpenStack in another room and proceed forward, essentially leaving VMware in stasis. But when OpenStack’s complexity set-in, so did the disillusionment. And there was something more. These users started to conclude that, even if they overcame the complexity, staffing and ongoing support issues, OpenStack wasn’t actually going to deliver the sought-after cost savings. They turned back to cloud-transforming VMware vSphere.
Leaders of the movement speaking here at the OpenStack Summit 2017 are clearly aware of at least some of the issues. In yesterday’s opening keynote, Mark Collier, COO of the OpenStack Foundation presented a one-word slide to the audience – Complexity. That was followed by another one-word slide: NIH (Not Invented Here). He conceded that, in the process of building a cloud development, deployment, and management model, the OpenStack foundation believed that it had to encompass the entire cloud code “stack.” Projects were added at a rapid pace forcing the need for equally rapid release cycles. Complexity grew exponentially. Collier concluded that the Foundation should now break away from its sense of cloud nationalism and reach out to other open source projects that could do a better job in specific areas. He called this “composability.” Ironically, his presentation was followed by an on-stage demonstration of OpenStack Cinder that “almost worked.”
Composability will allow users to build OpenStack deployments using best of breed open source components. It will (hopefully) break the NIH mentality. But at this point, I doubt that the result will be less complexity. As another keynote speaker, Brian Stevens, CTO of Google Cloud pointed out, it’s one thing to compose, but there is as yet no standard way to do this. There is now the need he said to create composed clouds in a standard way. This is certainly a standard well worth pursuing, but will composability alone be enough to deliver cost savings and bring disgruntled early adopters back into the fold?