Why the private cloud has stalled

The appeal of making your own data center cloudlike is easy to understand: Who wouldn’t want an entirely flexible commodity infrastructure, where you can pour on compute, storage, and network resources as needed?

From the early days of the cloud, CIOs saw promise in the cloud’s magic combination of reduced cost and vastly greater agility and wanted to capitalize on cloud architecture in their own organizations. But those closer to the ground in enterprise IT have always had a tendency to roll their eyes. They see the advantage of scale-out, self-service architectures for some applications — such as dev and test — but the return on investment in other areas is not always so obvious.

[ Learn all about the private cloud — the most important trend in IT — with InfoWorld’s Private Cloud Digital Spotlight. Download it today! | Stay on top of the state of the cloud with the “Cloud Computing Deep Dive” special report. ]

Part of that resistance is the usual allergic reaction to change, because major shakeups mean fewer resources to meet project deadlines and keep legacy systems humming. But there are also very good reasons why the private cloud is taking so long to take hold. The two leading ones are the immaturity and expense of available private cloud solutions.

Private IaaS shuffles along

The three major IaaS solutions remain the OpenStack framework– a growing collection of open source projects revised twice a year — and commercial IaaS solutions from Microsoft and VMware.

The word from the latest OpenStack conference is that adoption appears to have slowed, in part because OpenStack remains a moving target. To deploy OpenStack, customers generally need packaged versions — from the likes of HP, Rackspace, RedHat, or Ubuntu — or major professional services assistance. Both are often required by those serious about deploying OpenStack, and in the process, customers run a serious risk of being locked into the specific implementation they’ve adopted.

Microsoft has an interesting DIY hybrid cloud solution, where Windows Server and System Center together provide both private cloud infrastructure and a bridge to the Azure public cloud, but it’s still evolving. With its vCloud Suite, VMware provides the most mature private cloud solution, but it’s expensive enough to make you think twice about scaling it out.

From what I’ve heard, for the most part, private IaaS is still mainly used for dev and test — so developers can access self-service features to reconfigure their environments (or at least have ops do that more easily). Interestingly, if you add deployment, this is also what PaaS (platform as a service) was designed to do.

Private PaaS seeks momentum

Although most people think of PaaS in terms of such public cloud offerings as Amazon Elastic Beanstock or Heroku, I actually think private PaaS offers the most compelling value proposition for the private cloud.

Private PaaS provides a modern replacement for the application server. You develop, test, and deploy applications that scale out on commodity servers rather than scale up on premium hardware; also, you can accommodate multiple languages and (theoretically) wrap governance around application development across your organization. A scale-out platform that can embrace almost all applications you develop in-house provides the most compelling reason I can think of for building your own cloud.

Currently, there are two leading solutions: Red Hat OpenShift and Pivotal CF. Both have their pluses and minuses; OpenShift supports the widest swath of languages, but InfoWorld’s Andrew Oliver has noted a certain clunkiness in using the product. The “CF” in Pivotal CF stands for the open source Cloud Foundry project, which is gaining popularity, but the fully supported Pivotal CF product runs only on VMware — again, an expensive proposition.

There are lesser-known solutions, such as Apprenda (which targets .Net), Jelastic (based on Virtuozzo), and CenturyLink AppFog (based on Cloud Foundry). I’ve also heard good initial reports about IBM BlueMix, also based on Cloud Foundry, but it’s still in beta. In short, private PaaS is still very much at an early phase.

The Catch-22

The common-denominator benefit of the private cloud is the move to commodity scale-out infrastructure. That sounds fantastic — but it’s not what most enterprises have in place. Bending your existing infrastructure to accommodate a vast private cloud, not to mention migrating the lion’s share of legacy applications to that cloud, is a losing proposition.

Ultimately, building a private cloud demands replacing old systems with new ones. That can be accomplished step by step, as new initiatives — particularly those supporting customer engagement — demand cloudlike infrastructure. In addition, as legacy systems become too awkward and expensive to maintain, you can replace them with modern cloud solutions.

But even an incremental approach is difficult when, as I’ve suggested, the private cloud is in such flux. Can anyone tell at this point which private cloud solutions will have the most staying power? Also note that I haven’t touched on the subject of SDN (software-defined networking), which is still struggling to deliver solutions enterprise customers can deploy, yet is considered vital to making the most of cloud infrastructure.

The fact is that the major public cloud service providers have solved these problems. That, after all, is their charter as inventors and purveyors of the cloud. Yes, large enterprises that have the resources — not to mention major regulatory compliance constraints — will eventually build their own private clouds. Some already have. And of course, startups that both plan to scale quickly and maintain their own infrastructure tend to commit to cloud architecture from the ground up.

An established enterprise shifting to that paradigm is an altogether different proposition. The ongoing, relentless growth of Amazon Web Services may be the clearest indication that many will continue to avoid the pain, disruption, and expense of building their own clouds and, instead, embrace the public cloud.

This article, “Why the private cloud has stalled,” originally appeared at InfoWorld.com. Read more of Eric Knorr’s Modernizing IT blog. And for the latest business technology news, follow InfoWorld on Twitter.