OpenStack is the industry's most successful private cloud technology. That may not be enough.
As Gartner analyst Alan Waite posits, there are a limited number of use cases for which OpenStack actually makes sense. This isn't to suggest that OpenStack is DOA, but rather that the OpenStack community needs to find ways to work more closely with leading public clouds.
In other words, OpenStack needs to embrace Amazon Web Services. The problem, however, is that AWS doesn't need OpenStack nearly as much as OpenStack needs it.
Private or public?
That isn't soley my opinion. OpenStack pioneer and Cloudscaling CEO Randy Bias, analyzing AWS numbers, declares it's time to stop pretending the public cloud won't thrive and instead find ways to work with it.
Given that AWS has clearly "built something fundamentally new" that "is growing at an unprecedented pace," Bias insists:
It means we need to get serious about public and hybrid cloud. It means that OpenStack needs to view AWS as a partner and that we need to get serious about the AWS APIs. It means we should also be looking closely at the Azure APIs, given it appears to be the second runner-up.
The problem with this view, as OpenStack PR lead Robert Cathey suggests: It's very likely that "AWS doesn't see OpenStack as a potential partner," even if the OpenStack crowd finally recognizes the need to work with AWS.
In short, AWS is sticking to its public cloud guns, refusing to bend on the idea that private cloud is no cloud at all, as AWS chief Andy Jassy once opined. In the AWS mind-set, "Organizations that have private cloud systems will have missed out on all the advantages and benefits of going into the cloud."
Some will point to allegedly better pricing for private cloud, as 451 Research recently tried to unpack. But the reality is that private cloud pricing can be as opaque as server-based pricing has long been, and price isn't really the primary factor in public cloud adoption.
Private cloud's last great hope
Not surprisingly, the private cloud vendors don't agree. Or should I say "hybrid" cloud vendors?
For many companies, "hybrid" is the new "private" cloud, as private cloud has failed to deliver on its promises. (By one Gartner estimate, 95 percent of private cloud deployments fail for a variety of reasons.)
"Hybrid" has become the rallying cry for vendors ill-prepared to compete with the likes of Amazon and Microsoft, and for companies too wedded to the idea of their own private data centers to take off the cloud training wheels and move to the public cloud.
But OpenStack, argues Gartner analyst Alan Waite, is "our current best hope for a private workload control layer able to operate across technology silos." In other words, OpenStack can provide "a control plane and API for compute in much the same way that software-defined networking and software-defined storage are beginning to do for networks and storage."
That sounds good. That sounds valuable. It also sounds, he warns, like a very (very) incomplete answer to AWS and the public cloud.
OpenStack deployments, he cautions, number in the hundreds, not the thousands one might expect given the size of its community and hype, largely because of the "limited use cases where an OpenStack cloud would actually be the right solution."
As such, he concludes, "Anyone who thinks that the obvious next step for all server virtualization users will be OpenStack is sadly mistaken; the opportunity is much smaller than that."
That opportunity could be much bigger if OpenStack served as a partner and gateway to the public cloud. I've been arguing for years that OpenStack needs a dominant vendor, someone who can bring order to its community chaos. Red Hat did this for Linux, and Red Hat is well positioned to do the same for OpenStack.
Indeed, OpenStack is starting to pay serious dividends for Red Hat. In Red Hat's most recent earnings call, CEO Jim Whitehurst noted that half of the company's OpenStack deals topped $100,000, with a significant number of its top 30 deals in the quarter involving OpenStack (up three-fold since the quarter before).
There is clearly opportunity for OpenStack, even if the public cloud is growing at a considerably faster rate. As Gartner analyst Thomas Bittman highlights, "New stuff [workloads] tends to go to the public cloud ... and new stuff is simply growing faster" than more traditional data center workloads.
How much faster? Try three-fold.
Suddenly, Eucalyptus' API compatibility with AWS looks like a very smart move. HP acquired Eucalyptus last September, and we haven't heard much about its compatibility since, but for OpenStack to really hit its stride, it'll have to play nicely with AWS even if AWS doesn't care to accommodate it.
Sophisticated companies like Walmart are finding great success with OpenStack, but they're still the exceptions, not the rule. For OpenStack to be more than a safety blanket for cloud-shy CIOs, it'll need to find ways to work with Amazon and Microsoft.
As mentioned, Red Hat may provide an answer. Early on Red Hat made it easy to run Red Hat Enterprise Linux across different environments, from a corporate data center to AWS. The company has also extended its management services to make it easy to manage workloads across different deployment targets. That seamless experience needs to be replicated for OpenStack.
Customers will increasingly demand it, though more of that burden will fall on Red Hat and its ilk than on AWS. As Red Hat executive Paul Cormier told me:
The problem with the other PaaS providers is that if you build on any of the other PaaS solutions, whether it comes from Google or someone else, you're not getting the application out of their network. Ever. The app is only going to run on their network. Customers don't want this. They want to own the future of their application. Red Hat gives customers this ability by providing a consistent platform across different deployment targets.
Fortunately, Red Hat is already messaging the need to make OpenStack work with AWS. But now it needs to turn that message into reality.