Alistair Croll has an interesting post on gigaom’s refresh the net about understanding the various types of cloud computing that’s worth a read. He tries to break down cloud computing along two axis, whether you get to decide what software your run or the service provider does, and where the resources are located. He ends up with two classifications, development clouds where the provider selects the stack, and operations clouds where you select your software.

IMHO, though, I think Alistair has fallen into a trap laid for him by dozens of other bloggers and vendors - accepting the idea that anything run outside your data center is cloud computing. This notion, which started with folks relabling SaaS as cloud computing, eventually lead to the explotion of XaaS acronyms.

I have to admit, it’s been easy to do fall into this pattern, and I’ve even caught myself doing it. At the root of this confusion I believe is that many new cloud computing services cropping up today are really built from old infrastructure. On one hand, several companies have tried to copy EC2 by offering virtual machines provisioned through an api. On the other hand many services have cropped up offering hosted platforms, essentially shared software stacks deployed as a cluster.

With the proper technology, this type of tradeoff isn’t needed. Agathon Group, a new data center partner of 3tera, is using AppLogic to offer not only both virtual machines, and prebuilt software stacks, but also full virtual data centers. And, of course, they’re doing all this from a single physical infrastructure.

For cloud computing to truly succeed, requires real technology innovation. As new services come out the difference will become clear and the terminology confusion we’re experiencing today will subside.