A number of basic facts about the cloud are coming into focus now that deployments are moving beyond the test-bed stage.
First, the cloud is not like data center architectures of the past in that they tend to incorporate a much broader set of systems and technologies that in-house enterprise techs may not be familiar with. They also require a high degree of automation and system interoperability across a much more diverse landscape than even the largest of enterprises have had to deal with.
It is for these and other reasons that the push is on to build as much openness into basic underlying cloud infrastructure at this early stage of the game, a movement that has drawn key users, service providers and, most importantly, vendors. The prime example of this broad cooperation is Intel's Cloud Builder program, an initiative the company has undertaken in conjunction with its work with the Open Data Center Alliance. The program has cobbled together a who's who of IT technology leaders, including Cisco, HP, IBM, EMC, VMware, Citrix and Microsoft, in an attempt to establish an overarching framework for cloud interoperability and federation.
As Jason Waxman, Intel's general manager of High Density Computing, explains it, the cloud has the potential to produce 20 percent annual growth in IT infrastructure over the next half decade, which represents not only a lot of new silicon-using devices, but also the management, virtualization and application software to make use of it.
To that end, the group has launched an aggressive program to spell out the necessary reference architectures that will help system architects and integrators build fully federated cloud infrastructure so that no matter whose technology a given data set encounters on the cloud, accessibility, availability and overall user experiences are not compromised. At the moment, the group has published about 25 RA's-governing topics like building a basic vCloud environment to incorporating that environment into, say, Dell's power management stack and then creating pools of trusted hardware on which to run it. No one is really sure how many RA's will ultimately be needed, although Bill Cox, director of the Cloud Builder program, says it could easily stretch into the hundreds.
For Intel, the purpose of the effort is to allow as much flexibility as possible so users and providers can customize their own environments even while the overall cloud architecture enables or enhances key IT goals like improved automation, greater efficiency, unified networking and scale-out shared storage.
It's a lofty goal, nothing less than universal compatibility of all-things-cloud. But the simple fact is that without it, the cloud is likely to wind up in the dustbin of history as yet another failed attempt at open IT environments.
At the moment, the desire is in place to make it happen, and the first set of reference architectures are a good sign that execution will not fall prey to internecine squabbles. But technical challenges remain, and whether the group can produce a truly federated cloud ... well, that's the million (or billion) dollar question.