Defining cloud computing

I liked this post by Larry Dignan on the cloud computing buzzword and how meaningless it has become.

Writing on the subject recently, I was struck by the gulf between what some people mean – online apps like Google Apps and Gmail – and what others mean, on-demand utility computing such as that delivered by Amazon Elastic Compute Cloud or Flexiscale. These things have little in common.

Dignan has even more examples.

Should we abandon the term? Maybe, but I find it useful if only as shorthand for describing how the centre of gravity is shifting to the Internet.

Some services are more cloudy than others. Dignan refers to this Forrester report (though you’ll have to look at the blog post for the extracts, unless you want to buy it) which has a table of “six key characteristics.” I don’t agree with all of them; the business model, for example, is not an inherent part of cloud computing. I am interested in number two:

Accessible via Internet protocols from any computer

Any computer? OK, probably not the Atari ST which I have in the loft. Any computer with a web browser? What about requiring a “modern” web browser, is that OK? Java? Flash? Silverlight? A specific version of Java or Flash? What about when we need a runtime like Adobe AIR or Microsoft Live Mesh? What if it doesn’t run on Linux? Or on an Apple iPhone? What about when there is an offline component such as Google Gears? All these things narrow what is meant by “any computer”.

This is the old “rich versus reach” debate; it is still being played out. My point: cloud computing isn’t a boolean characteristic, but a continuum from very cloudy (NTP) to not cloudy at all (Microsoft Office).

One thought on “Defining cloud computing”

Comments are closed.