And with that, OpenStack is now unquestionably a big vendor driven set of projects. EMC acquired CloudScaling a while back, Cisco has announce their acquisition of Piston Cloud, and IBM is acquiring Bluebox. The only meaningful independent OpenStack generalist company now is Mirantis. (props to HP, RedHat, and Canonical, but they also do other things).
It’s not that anyone really ever questioned that OpenStack was being driven by corporate interests. The cliche has always been that it has more vendor sponsors than customers. But does that matter? The point of OpenStack seeks to provide a common IaaS layer that’s not owned by Amazon, so that all these corporate interests can collectively catch up to the head start Amazon enjoys. A the same time, corporations that feel like hosting their own IaaS is strategic to their business are encouraged to consider OpenStack since their traditional IT vendors are also leveraging it as an emerging standard.
What are these traditional players going to do with these OpenStack upstarts? ensure compatibility with existing solutions… build OpenStack-in-a-box products… provide service and support offerings around the platform… make sure that there’s just enough innovation within OpenStack solutions to be competitive, but not too much that would devalue existing products too quickly… you know, the standard stuff.
From the Vancouver Summit, though there appear to be more direct customers using “OpenStack,” it’s more nuanced than that. The nuance is OpenStack is not a product or a single project. OpenStack is a collection of projects that encompass compute, networking, and storage. Customers do not have to swallow the whole pill. Many of the customers “using OpenStack” at the summit are really only using Nova, Glance, and Cinder; or Swift; or Ceph (not OpenStack BTW); and very few are leveraging most or all the projects for an all-encompassing deployment.
I think OpenStack has a future. It’ll be up to the governance model to ensure that OpenStack remains a common playing field or diverges into separate incompatible offerings. It’ll be fun to watch the run-by-committee model and see if it can produce a truly viable IaaS before while such a thing is still a relevant need in IT.