OpenStack turns two this week. That means the open-source project — which fancies itself the Linux of the cloud — is entering a critical stage of its development process.
Rackspace — which helped give birth to OpenStack in July 2010 — rolled out some stats to show OpenStack momentum and to push its OpenStack-as-Linux comparision. For example, in the 84th week of the project, there were 166 entities contributing to the effort whereas it took Linux 828 weeks to hit 180 active contributors, according to Rackspace’s tally.
By that count — and the fact that Internap, Hewlett-Packard, and Rackspace itself have OpenStack clouds in some degree of production — shows pretty fair momentum going into year three. Whether that’s enough to form a competitive counterweight to Amazon Web Services and VMware vCloud’s heft remains to be seen. And some competitors (most notably Eucalyptus’ CEO Marten Mickos) maintain that OpenStack has too many cooks in the kitchen. At GigaOM Structure last month, Mickos said OpenStack, rather than following in Linux’ footsteps, could become “the Unix of cloud.” The implication was that so many vendors weighing in could lead to a forking or fracturing of the OpenStack standard.
Fear of fracturing
The project is transitioning from an effort largely driven by Rackspace to a broader OpenStack Foundation with control to be spread among several vendors including Platinum members AT&T, Canonical, HP, IBM, Red Hat and SUSE joining Rackspace. (Gold members include Cisco, ClearPath Networks, Cloudscaling, Dell, Dreamhost, Morphlabs, NetApp and Yahoo). There is precedent for this model succeeding. The Eclipse Java IDE really took off once IBM ceded control to a multi-vendor foundation.
Jim Curry, GM and VP of Rackspace Cloud Builder, one of the early instigators of the OpenStack effort, is clearly bullish on the project’s prospects but acknowledges its challenges as well. “Any open-source project that is not controlled by one vendor but with lots of folks with their own agendas is at higher risk of fracturing — the advantage of single-vendor or close-to-single-vendor projects is that there is more control in the short term and a shorter time to market,” he told me in a recent interview.
But, in his view, that dynamic shifts over time. “For continuing, rapid innovation it’s up to the community to determine if standards, interoperability are important. It’s hard to make that judgement now because OpenStack is still really early.”
There has already been agita among the OpenStack alliance with Citrix, an OpenStack backer, reversing course in April, setting up its more mature CloudStack as a rival open source cloud platform. Now there is concern in some quarters that even some of the Platinum partners are hedging their bets with OpenStack, while pursuing their own potentially competitive projects. Red Hat’s CloudForms is one example. (For the record, Red Hat paints CloudForms as complementary, not competitive, with OpenStack although some of its own partners said there is deep confusion on that point among customers evaluating cloud options.)
Curry said there have been some vendors — although he stressed that he does not count either HP nor Red Hat among them — that have used the OpenStack brand for its marketing value while not fully buying into the mission. “Red Hat and its CTO Brian Stevens are big backers and they’ve contributed a lot of code but large companies like that often have competing solutions. Some of them launched before OpenStack, but I love the contributions from those companies and others.”
Bottom line: Choice is good
Curry holds true to OpenStack’s early mantra: customers need choice in their cloud deployment options. “Cloud was not this big two years ago. At that time, customers had to choose one technology or another, the choices being Amazon and VMware. It’ s not that they were bad technologies — they’re great technologies with a lot to offer but by choosing one of those you limit the opportunity to go to another,” Curry said
As OpenStack enters its third year — and as Amazon and VMware face customer fears of vendor lock-in — everyone agrees that it’s still early in the cloud game. Most companies have put some projects in the cloud but have not yet fully bought in. But they all want choices.