HomeCloud

Outlook partly Cloudy

Like Tweet Pin it Share Share Email

The National Institute of Standards and Technology (NIST) is on my mind. I’ll get to why in a moment. Right now I have a project for it to take on: Public Infrastructure.

Here’s why: Right now public discussions of political issues such as taxation are frequent, top-of-mind, polarized, and controversial. They are also entirely divorced from the largely apolitical issue of maintaining the public infrastructure – a subject which is, in the minds of most Americans, foggy.

To illustrate: See if you can, in one minute or less, make a list of just ten elements of the public infrastructure whose absence would seriously impair our economy or way of life.

This matters because those who polarize us depend on our failure to recognize how much we all agree. For example, few citizens consider the Interstate Highway System to be an area where private ownership makes sense; likewise sewage treatment plants; the police and fire departments; and the courts, Judge Judy notwithstanding (add them all, except for Judge Judy, to your list).

With such a list … not of just ten elements but of everything that’s beyond controversy, along with the itemized costs of maintaining them all … we would all know the minimum total tax revenue required to maintain our society, and the tax debate would be far less noisy.

NIST should be on your list (please forgive the long lead — the actual topic is close at hand). Relatively few Americans have any idea that there even is such an agency. It isn’t indispensable in any theoretical way.

But while it might be fun to watch the business community collectively plotz as its members suddenly had to pay for the colossal variety of services NIST provides, I personally would prefer to avoid watching the DOW collapse from the weight of the financial burden its elimination would impose.

What brought this all to mind was a note from subscriber Mike Gautier last week in response to last week’s tirade about the Cloud, and how its proponents have turned its definition into “everything.” Mr. Gautier pointed me to a NIST publication titled “Draft NIST Working Definition of Cloud Computing v14,” by Peter Mell and Tim Grance, which is far and away the clearest account of what the Cloud is and isn’t that I’ve ever seen. Since NIST is a no-nonsense agency that maintains public standards, this is, while refreshing, not entirely surprising.

Mell and Grance define the Cloud as follows:

Cloud computing is a model for enabling convenient, on-demand network access to a shared pool of configurable computing resources (e.g., networks, servers, storage, applications, and services) that can be rapidly provisioned and released with minimal management effort or service provider interaction. This cloud model promotes availability and is composed of five essential characteristics, three delivery models, and four deployment models.

The rest of the two-page document explains the five characteristics, three delivery models, and four deployment models. Every IT leader should be familiar with them as a matter of basic literacy in matters of technical architecture.

Regular readers of Keep the Joint Running are aware of my skepticism regarding the Cloud, the result of so much breathless over-hyping by so many naive-but-influential enthusiasts.

The electrical-generation-model, where IT will consume computing cycles and storage without knowing where they come from, is utter nonsense. If you do know where they come from … Amazon.com or Google, perhaps … it’s data-center outsourcing if you pay, and risky if you don’t.

Meanwhile, Software as a Service will continue to mean “hosted solutions,” and its integration challenges will continue to outweigh its utility for most applications.

Elastic infrastructure provisioning (EIP anyone?), on the other hand, is a meat-and-potatoes capability you’ll want as soon as you can get it, whether it’s inside your own data center or provided by a data-center outsourcer, because the benefit is clear and disruption is limited.

Right now you forecast capacity requirements separately for each and every application in your portfolio, and provision infrastructure accordingly. With EIP, in contrast, you would only have to forecast capacity requirements in the aggregate. All applications would draw computing cycles and storage out of a shared pool. The economies would be immediate because you’d need less total capacity.

Maybe it’s okay that the industry’s naive-but-high-profile visionaries include EIP in their account of “The Cloud.”

For us pragmatists, though, EIP is “merely” the merger of virtualization and load-balancing, taken to a higher level.

Maybe that’s why we don’t get the big buck speaking fees.