Platform Computing: Herding Clouds

After the post Platform Computing Announces Infrastructure Sharing Services appeared on ZDnet, I had the opportunity to speak with Songnian Zhou, CEO of Platform Computing. Although the plans were for me to speak with Platform prior to the post going live, scheduling problems made that impossible. So, the focus of our conversation moved from the announcement of Platform's new product to a more general, philosophical review of the industry march towards the adoption of cloud computing in some form.

Here are some important points from our discussion:

• Cloud Computing is in the "over hype" stage of development. That is, anyone doing anything vaguely touching on distributed, virtualized computing that offers pay-as-you-play pricing and self management is waving the cloud banner. The industry must find a standardized definition of what is and isn't cloud computing. My colleagues at the 451 Group, Rachel Chalmers, William Fellows, Dan Golding, Antonio Piraino and others have done their best to do just that in an upcoming report.
• Management is one of the key requirements and features of a complete cloud computing environment. The good Dr. Zhou would point out that Platform has been offering powerful management tools for distributed, extreme computing environment for well over a decade and is well positioned for the evolution into cloud computing.
• While it appears that industry standard systems are an increasingly important part of organizations' overall computing strategies, cloud computing is likely to include products from suppliers of many systems, operating systems, virtualization tools, databases, development tools and applications. Management tools that only support a single product in any of those categories is likely to be of limited use to many organizations.
• Giving organizations the ability to gently move workloads from their own IT infrastructure to that offered by an outside supplier and back again upon need is a critical requirement.
• Being able to control the environment so that costs and execution location are always known, is also a critical success factor. No run-away processes should be allowed.