The cloud is the greatest technical achievement of a generation, poised to deliver operational advantages and cost savings that would have been inconceivable only a few short years ago.
And yet, there is growing concern that the IT industry, and world at large, does not fully comprehend all the ramifications of this monster that has been created. After all, the cloud would not be the first scientific development to show tremendous promise at first, only to be followed by serious, intractable problems.
The sheer flexibility that is hardwired into cloud architectures could prove particularly nettlesome, according to Yale researcher Bryan Ford. As users offload responsibilities to cloud providers, who then transfer it to ever widening circles of alternate providers, the potential exists for data to be spread across a veritable stew of infrastructures, architectures and hardware and software resource pools. The interactions between these elements will prove particularly difficult to manage and could ultimately cause widespread meltdown of data environments. Think of it as the IT equivalent of the financial derivatives market that even so-called experts failed to understand until the entire world economy was on the brink.
The ease at which individual business units can spin up their own cloud resources, often leaving IT out of the loop entirely, should be a key cause for concern, says E-Commerce Times' Jeffrey M. Kaplan. The cloud will usually provide a low-cost alternative to internal resources, but the cost will mount over time. Since they are not well-versed in the nuances of enterprise data architectures, most business managers fail to realize their mistake until it's too late, saddling the organization with either a steadily increasing bill for services or a hefty expansion of O&O infrastructure. One rule of thumb: Utilize the cloud primarily for short-term projects.
Still other downsides to the cloud are less tangible than costs. Loss of control is a major factor that many enterprises do not fully comprehend in the heady rush to keep up with the times. As theserverside.com's Cameron McKenzie points out, a certain level of control can be had with IaaS architectures, but be wary of some 'IaaS Plus' configurations in which the service provider adds his own app servers, database management, middleware and other components. This puts a fair amount of control in the provider's hands - things like container configuration, JVM provisioning and port assignments - which ultimately may not deliver the optimal environment you think you are paying for.
Clearly, cloud computing calls for an entirely new mindset among data technologists, says GigaOm's James Urquhart. The challenges of managing physical resources fostered "reductionist thinking" that enabled problem-solving by isolating and solving root cause problems. On the cloud, what's needed is "systems thinking" in which knowledge of entire environments leads to greater transparency and more effective operational acumen. The significant challenge, however, will be to define a given cloud environment's boundaries considering the fluid nature of its operation.
So is IT heading down the same road as Dr. Frankenstein, driven by dreams of the betterment of mankind but ultimately doomed by his own creation? That may be a little over-dramatic, of course, but it is the rare CIO who has not encountered the law of unintended consequences when trying out advanced architectures.