A year ago this month, it was abundantly clear that both the OpenStack project and the foundation created around it were at a crossroads. After years of a “Big Tent” strategy that left the project with members trying to take the project in different directions and reinventing various wheels, a strategy that made it hard to even answer the basic question of what OpenStack was, last year’s annual gathering of OpenStack project adherents was a step in a new direction. From the aggressive embrace of Kubernetes to the upleveling of projects distinct from OpenStack itself like Kata Containers and Zuul, the OpenStack Foundation was clearly signaling a willingness to break with its past, to chart a new direction for itself.
The question heading into this year’s Summit, held in beautiful Denver, Colorado, was whether the foundation would continue down this new path or regress back towards its own history.
The answer was evident everywhere one looked at the conference, perhaps nowhere more so than with the event’s branding. What had previously been known as the OpenStack Summit was reborn as the Open Infrastructure Summit, a name that carries forward the original mission while simultaneously opening its doors for an array of new projects. If it’s true that the OpenStack Foundation was at a crossroads a year ago, it’s equally clear that it has picked its path moving forward.
The obvious question for many in the industry is: does this matter? For all of the project’s grand ambitions nearly a decade ago, the conventional wisdom today is that the clouds that matter are the ones operated by entities willing to invest double digit billions per year in infrastructure. As with all technologies that ride the hype cycle, OpenStack hit its trough of disillusionment and for years it’s been something of an industry sport to kick the project around.
But that doesn’t answer the question of whether the project matters. Certainly it does to users such as Verizon Media with its four million cores running on top of the technology. But are such cases outliers, or does OpenStack have a broader opportunity in front of it?
This is a common question asked about the project, maybe the most common. Fortunately, it can be answered simply depending on the answers to two corresponding questions.
When members of the press ask whether OpenStack has a sustainable pathway moving forward, we ask two questions in response. First, do you believe that all on premises infrastructure will go away in favor of public cloud infrastructure? And second, do you believe that products like the vCloud/vSphere combination matter?
It is common in the industry to see arguments that public cloud infrastructure is the only reasonable approach. Given the economies of scale involved that permit large hyperscale providers to purchase everything from hardware to power to network access more cost efficiently than any other enterprise regardless of size, given the hyperscale providers ability to attract elite talent, given the network effects of large product and service portfolios with each enhancing the value and decreasing the friction of adoption of the next, the arguments on paper that most if not all workloads should move to the cloud are clear.
The difficulty with these arguments, however, is that the enterprise is not now, and never has been, run by arguments on paper. If they were, enterprises would have declined to pay the Office and Windows taxes years ago when open source or SaaS-based alternatives became viable for the majority of basic use cases. Assuming that all or some close approximation of workloads will move to the cloud because of the obvious case on paper for such a move assumes that all enterprise decisions are made by strictly rational actors.
For anyone that has spent any time in enterprise procurement or studying the history of it, this will seem like a problematic assumption. Which leads to our first answer, which is that we are likely to have on premise infrastructure for the foreseeable future. This is certainly the view of the public clouds, in fact, as evidenced by offerings such as AWS Outpost, Azure Stack or Google Anthos.
This answer, in turn, answers our second question. If one assumes that on premise infrastructure will remain indefinitely, even if it will not match the growth rates in the public cloud, then clearly the vCloud/vSphere product line from VMware is an important one. Because having been exposed to the instant elasticity and dynamic provisioning of the public cloud, enterprises are unlikely to accept private infrastructure that does not at least approximate the ease of use and accessibility of the public cloud.
If one is going to argue that OpenStack has no opportunity in front of it, then, it is necessary to explain either why there will be no private infrastructure or why in a world in which private infrastructure remains a viable option that OpenStack will not be a candidate for those workloads.
Much depends on execution against that opportunity, of course, and OpenStack’s track record in this regard is mixed at best – particularly the further back in the project’s history one goes. But for all of the flak that the project takes, OpenStack already has large customers running sizable workloads on its technologies. More importantly, its new, inclusive approach moving forward certainly has the potential to pay dividends.
It’s interesting, in fact, to consider what the Foundation could become. While the project was once intended to conquer the world on its own, its implicit acknowledgement that it will take more than just an IaaS software layer to get there hints at wider possibilities. This year’s rebranding to the Open Infrastructure Summit was more than just a name change, clearly; projects like Airship, Kata Containers, StarlingX and Zuul were given near equal billing both on the mainstage and in analyst sessions.
But the name also begs broader questions: what would an open source alternative to re:Invent look like? What if the OpenStack foundation was able to ally itself more directly with iniatitives such as the CNCF and the Open Compute Project?
It’s likely too early in the OpenStack Foundation’s pivot away from its Big Tent heritage to be asking these questions in earnest. As someone who believes that the divisions between open source projects represent the industry’s biggest limiting factor in competing with the likes of Amazon, however, such questions probably should be asked and answered sooner rather than later.
Disclosure: The OpenStack Foundation is a RedMonk client and paid for T&E to the event. Amazon, Google, Microsoft and VMware are also RedMonk customers. The CNCF and Open Compute Project are not RedMonk customers.