Wednesday, May 14, 2014

Reconsidering Cloud Computing

I've spent quite a bit of time with my eyes and ears open, watching and listening to expert practitioners in my professional network talk about, and work on, cloud computing projects.  A reconsideration, and a sharpening, of my position is in order.

As readers know, I have been very skeptical about the whole catch phrase called "cloud computing" since its inception, which seems to have been over six years ago, if one were to take serious discussions in professional IT forums.  Dr. Irving Wladawksy-Berger is a retired IBM scientist who writes a rich blog on computing topics, with lots of links that allow me, as a primary source, research-driven type, to delve into some of the seminal research publications in this field.

As Dr. W-B points out, most of the meetings from five or six years ago began with the question, "What Is Cloud Computing?"  The various, vague, contradictory answers given, along with the obvious self-interest of the consulting groups, made me comfortable that it was hype, full of sound and fury, signifying nothing of lasting value.  Today, he suggests that Cloud is a new model of computing, but that there is no single dimension to define a computing model.  Taken together, this means that Cloud, like pornography, may a priori resist definition, but we'll know when we see it. 

An IBM glossy publication says, "Cloud computing is a pay-per-use consumption and delivery model that enables real-time delivery of configurable computing resources (for example, networks, servers, storage, applications, services)."  The term pay-per-use wasn't adeptly chosen, but you get the idea.  It is a workable definition from which to start.

IT consulting companies, such as Accenture, like to trumpet lots of advantages for cloud computing, but the ones that don't require using IT to become a "disrupter" are cost flexibility and scalability.  The cost flexibility comes from taking large fixed costs of buying computing power (servers, or even supercomputers), storage, data analytic software, applications, development tools, and services up front and either capitalizing or expensing them; a preferred vendor could provide them customized and bundled on a metered, or variable cost basis. The second benefit, related to this one, would be avoiding the lumpiness of buying a big system and growing into it.

We've always felt, from some of the linked posts above, that IBM would have a compelling offering in cloud computing.  We also suspected that HP could also be a player, once it had raised itself from its self-imposed, corporate death bed.  We are wondering about HP's role, beyond their many pronouncements.

As Dr. W-B writes about cloud computing, "The cute child in now a teenager, still full of promise, but also full of mischief if not properly handled."  This is  a very appropriate metaphor.

No comments: