Private Versus Public Cloud Computing
A plethora of applications are being considered for the cloud, but it may take at least another year before cloud computing goes mainstream in the enterprise.
Technology experts may argue over the relative merits of the private cloud versus the public cloud, but the fact is that even moderately sized enterprises are bent on turning their static data architectures into flexible, service-oriented ones.
The question remains, though: Will you know a private cloud when you see one?
To those whose job it is to develop cloud services for both internal and external infrastructures, private clouds are a piece of cake. LogicWorks' Keith Pollow has devised the following definition:
A private cloud computing platform's infrastructure is composed of a group of virtual servers with redundant technology, a dynamic computing network, and storage hardware dedicated to a selected business for the purpose of cloud computing.
Sounds simple, right? Of course, one definition begets another. How exactly does one build a dynamic computing network? And if the environment is so dynamic, why are we dedicating storage resources to any purpose at all, cloud-related or otherwise?
What about the plethora of cloud computing platforms out there? Certainly, once one of these is up and running, you've got the makings of a private cloud? Microsoft's new System Center 2012, for instance, has a new Virtual Machine Manager that essentially places a layer of abstraction on top of the already-abstracted virtual environment, providing for easy construction of fabrics, services and other cloud components. And open solutions like Cloudscaling's OpenStack-based OCS platform provides everything from management and security tools to hardware blueprints to turn compute, storage and network infrastructure into working clouds. Call them instant clouds - just add water and stir.
Most organizations, however, are already steeped in virtual technology. So where does the virtual layer end and the cloud layer begin? According to Virtualization Review's Elias Khnaser, key identifiers for cloud architectures include modular hardware scalability coupled with logical resource partitioning, service-based application architectures instead of traditional client-server-OS designs, plus healthy doses of self-service and automation.
Or could it be that we are approaching this all wrong? Instead of setting out to build a private cloud, maybe try defining present and future IT requirements first and then tapping into whatever technology produces the most satisfactory results at the lowest cost?
In this way, you deliver greater value to your organization, provide a higher degree of service to users and tap into technologies that actually meet short- and long-term needs. If someone says that's not really a cloud, tell them clouds are already obsolete.