by Greg Papadopoulos

The Beauty of Utility Computing

Feature
Dec 19, 20065 mins
Networking

When your business depends on a network of complex computer systems, just the thought of making dramatic changes can cause real pain. Even the promised panacea of utility computing—”as easy as buying electricity or phone service”—makes many CIOs break out in a cold sweat. They worry about the loss of control. Place the company’s data in someone else’s hands? Instant migraine.

This is where the proponents of utility computing—and I’m one of them—generally break out the money-in-your-mattress comparison. Is it safer there, we ask, or in a bank?

The obvious answer is that it’s not only safer in the bank but more accessible as well, thanks to a vast network of automatic teller machines.

It’s a compelling point, I think, but not the most compelling one.

The most compelling point is that utility computing is already a fact, and its widespread adoption will happen more quickly than most people think. Not convinced? Consider the banking example a little more closely. How did you feel about online banking when it was first introduced? OK, now, when was the last time you actually wrote a check?

Just as most of us now pay our bills online, many enterprises already make use of services where they don’t own or operate the infrastructure. Think e-mail, search, electronic storefronts and payment systems, customer relationship management.

Startups, in fact, are putting whole businesses together using Yahoo!, Google, eBay, Amazon or salesforce.com. The trend is also benefiting small to midsize businesses that don’t have the resources to build their own computing grids—and don’t have to.

Now, good CIOs pride themselves on the fact that information technology gives their companies an edge—and I have no doubt that it does—but consider where exactly that edge comes from. Most likely that won’t be in the base layers of the IT stack—the physical plant, virtualization and container (core services) layers. There’s plenty of innovation happening in those areas, especially in terms of energy efficiency, workload optimization and so on—and adopting those innovations brings real bottom-line advantages. No question. But the greatest efficiencies, and economies of scale, come with the utility model.

In other words, your compute-power company will be willing to spend more than you would for even a small gain in efficiency because the payback comes on a much broader scale. It’s simple economics. For example, the company I work for, Sun Microsystems, offers compute power at $1 per CPU/hour and data storage at $1 per GB/month on the Sun Grid—and I bet that’s a lot less than it costs you to do those things in-house.

(Note that I’m not talking about outsourcing, where you give someone your customized IT to run for you. That becomes very expensive over time and really slows you down because any changes you want to make have to go through contract negotiations. No, I’m talking about a true utility that offers a standardized infrastructure and, as a result, enjoys economies of scale you could never attain on your own. Again, think electric power or telecommunications services.)

Take advantage of those economics and you gain something else: focus.

Imagine being able to concentrate your energies on orchestrating the workflows that define your business and the “secret sauce” IT projects that differentiate your company from the competition.

That, to me, is the best argument for utility computing.

So, where to start? First, take stock of your current infrastructure. What are your cost centers? Your profit centers? The answers should give you a pretty good idea of where you want to be a consumer versus a creator of services. It’s really a question of where you can provide the most value. In other words, where are you using IT as a competitive weapon in support of your business?

Even in areas where you decide to own and operate your own services, think of them as utilities within your own shop. Build them on the patterns of the utility model. Then, if it later turns out that those services become less critical to you—or you find that you need to scale them in ways you hadn’t anticipated—you can move them out to a utility provider without having to re-architect everything.

Simply put, you’ll be keeping your options open.

In summary, my advice to CIOs is this: Let compute-power companies focus on the base layers of compute, storage and core-service composition. Consider that the substrate on which you can build your unique services, your enhanced customer relationships, your real competitive advantage.

Two things will happen. You’ll uplevel your job and your company.

Greg Papadopoulos is chief technology officer and executive vice president of research and development at Sun Microsystems.