Pushing the Limitations of Distributed Computing in the Cloud

January 10, 2012 Off By David
Grazed from IT Business Edge.  Author: Michael Vizard.

Enterprise IT continues to get more complex. It’s already difficult to manage distributed computing environments that now have as many, if not more, virtual servers than physical ones. And now IT organizations are wrestling with the prospect of having to manage all those servers across hybrid cloud computing environments.

IT vendors such as IBM in the case of the zEnterprise mainframe or Cisco in the form of the Unified Computing System (UCS) X86 server are responding to these concerns with homogenous platforms that allow IT organizations to consolidate the management of servers, storage and networking under a common management platform…

But those approaches require millions of dollars to actually implement. As a result, it might take years for IT organizations to centralize all their servers. In the meantime, the problem of managing widely distributed physical and virtual servers is only going to get more pronounced.

Convirture CEO Arsalan Farooq says that unless the IT industry as a whole finds some way to rise to this management challenge, the whole shift to cloud computing may be in jeopardy. IT organizations moving to the cloud will find that the management headaches associated with cloud computing are exponentially higher and many of them might simply declare cloud computing to be a failed experiment.

Convirture provides an open source framework for managing Xen and KVM virtual machine environments. Farooq says that one of the challenges that IT organizations are going to have to come to terms with is that not only is IT becoming more complex to manage, but the diversity of the virtual machine environment is increasing as rival platforms continue to gain share on VMware.

Potentially worse yet, Farooq notes that application workloads will need to be managed across increasingly fragmented enterprise IT environments. In an ideal world, Farooq says application workloads would be dynamically routed to the optimal platform available to process them. But he says IT system management vendors are still years away from making that happen across heterogeneous environments.

What all this means, says Farooq, is that the next two years may be a make-or-break time for a distributed computing model that is already pushing up against its own cloud computing limitations.