Now that a good number of enterprises have gained a modicum of experience with public cloud architectures, attention is turning in earnest toward replicating those environments on internal infrastructure.
The private cloud, in fact, is expected to be one of the chief growth areas for both enterprise-class hardware and software as organizations seek to first build the broad scalability needed to support a functioning cloud, and then the virtual and software layers to make it happen.
Indeed, the private cloud has emerged as a top priority within the enterprise vendor community as it provides a unique opportunity to remake the entire data infrastructure stack from the ground up. Dell, for example, has zeroed in on the private cloud now that its lengthy privatization process is complete, teaming up with Red Hat to integrate the OpenStack-friendly RHEL 6.5 across Dell’s data center portfolio. Dell will also take on RHEL service and support functions, even if the system is deployed on non-Dell hardware, a testament to the company’s desire to function within what is likely to be a broad, multi-vendor environment.
As well, HP has been busy talking up the private cloud, primarily as a means to integrate internal infrastructure with public services. But unlike Dell, which plans to work strictly with leading cloud providers like Amazon and Microsoft Azure for its hybrid environment, HP is also looking to build its own public component. The company recently announced several key upgrades to its CloudSystem platform, such as a new user interface that provides a single view of both on- and off-premise infrastructure and services. At the same time, however, the system is the first to run under the new HP Cloud OS, an OpenStack distribution that would allow users to also deploy cloud infrastructure on leading providers like Rackspace.
The notion of an all-encompassing, fully integrated public/private/hybrid cloud has held sway in enterprise circles for several years, but lately some experts are questioning the feasibility of such a system. In the first place, says technology author George Reese, public and private clouds are two very different animals, and anyone who thinks they will be able to mirror public cloud scalability and flexibility on private infrastructure is fooling themself. As well, the cost structures are very different, with private clouds requiring significant up-front capital expenditures and abandoning completely the idea that in the cloud you pay for only what you use. At the same time, private clouds are more at risk from power failures, natural disasters and other factors that actually make them less reliable and secure than public services. And if your ISP goes down, so does your cloud.
And it’s not like private clouds can be implemented at the drop of a hat, says Connectloud CEO Zeeshan Naseh. Aside from the cost to convert legacy infrastructure to cloud-ready status (or, barring that, building entirely new infrastructure from scratch), existing IT staff will have to surmount a substantial learning curve. And if you think building a private cloud on an open platform like OpenStack frees you from single-vendor dependence, thing again. Vendor reliance can be maintained in multiple ways on an open system, on the physical or virtual layers, or even higher up the stack.
Mind you, no one is saying that the private cloud is not a worthy endeavor. In terms of streamlining infrastructure, driving greater efficiency and creating a more flexible environment, the private cloud is the way to go. But it will not be an exact mirror of your public service, and the overall utility of bridging public and private infrastructure is still largely untested.
It’s a safe bet to say that “the cloud” will eventually coalesce into something resembling a cohesive data environment, but at this point the details are still very much up in the air.