John Wilkes:
Inventing the Storage Utility

Wilkes, who was recently named an HP Fellow, is explaining the vision behind his team's work.

"We want to give people more information, in a more digestible form and in a way that's more accessible," he adds.

One way to do that, Wilkes believes, is to be smarter about the way we create, configure and manage the data storage that is an essential part of any computer system.

self-managing storage
In particular, Wilkes and his team have been working on making large-scale, multi-terabyte online storage systems much easier to manage. Their idea is that the storage system should essentially manage itself.

This would mean that the people in charge of the system could concentrate on making business decisions, rather than continuously reacting to events from the storage system. It also would relieve them of the need to understand all the
tradeoffs inherent in these complex systems -- they could focus on what they wanted the system to do, rather than the details of how the task was achieved.

Wilkes and his team also see a time when users' data will follow them to whereever it's needed -- a concept they called "iShadow." It's slogan: "wherever you go, there's your data."

For example, if you travel from San Francisco to Boston to present complex information stored on your office computer, your data goes with you. When you arrive in Boston, your data has been automatically transferred to a nearby server.

trading storage capacity
Taking the concept a level deeper, Wilkes and his team now envision a world in which storage and computation resources are assets that companies trade. He sees computer systems negotiating those trades with each other automatically.

The same ideas can be applied to computation, he suggests. "You could imagine a situation where you didn't buy quite enough in-house capacity for a particular crunch you've got coming, but it turns out that one of your partner sites has just completed a compute-intensive design project and now they have spare capacity," Wilkes suggests. "Yes, they're in France, but the system can address all your data movement, security and trust concerns."

shared storage model
This vision of how storage will work in the future is substantially informed by one of the ideas for which Wilkes is best known -- automatic storage management. Wilkes views HP as being on a road to a future in which computer systems do more work for us by themselves, and need less care and feeding by people.

"Perhaps this is the defining thrust of my research," he says, pointing to the work of the storage systems group, which has been exploring this idea since 1995.

The complexity of modern computer and storage systems often threatens to overwhelm users. To address that, Wilkes led the development of the SNIA Shared Storage Model to bring some order and structure to the design space. He created the model in his capacity as a member of the Technical Council of the Storage Networking Industry Association (SNIA).

"The idea was to organize and describe a lot of the things that were taking place in the field," he explains.

The result was a way of thinking about storage that could provide a "map" of almost all the storage system designs being deployed, which in turn allowed vendors and buyers of storage products to see where they fit in the rapidly changing world of computer storage.

"The model's meant to be descriptive, not prescriptive," says Wilkes. "It's a tool. We've used it in explaining how what we do fits into the world we see, and we are starting to see both storage companies and customers doing the same."

doing world-class work
Wilkes' work with professional bodies like the SNIA and the Association for Computing Machinery (ACM) -- he was recently named an ACM Fellow -- has allowed him to keep in touch with the more academic side of computer research.

"One of the beauties of HP Labs," he says, "is you have a chance to have a foot in both camps. I also get to be an adjunct faculty at Carnegie Mellon University, which means I have an absolute blast working with these very, very bright people who are exploring wonderfully far-flung ideas."

It works the other way, too. Wilkes sees outside affiliations as a way to help HP Labs retain its status as a world-class research institution.

"Life is too short to be doing things that are not world class," he argues. "And how do you know if you are doing that kind of work? One way is to take advantage of the academic peer review process. So we aim to publish in places where the best people are publishing because we want to get that validation."

labs a good fit
This kind of attitude, and the high profile he enjoys within the storage industry, go a long way to explaining why HP recently decided to make John Wilkes an HP Fellow.

A Brit who came to HP's Palo Alto lab straight from his PhD work in Cambridge, England, Wilkes has just marked his 20th year with HP Labs. It's clearly a place that suits him well.

"I like HP's style," he says. "It has a corporate culture that basically values its people and it does what it can to try and embody that in the way that it does things."

"I'm not a great believer in hierarchies," Wilkes adds. Indeed, until he was made a Fellow, he never printed a title on his business card. "Labs happens to fit my slightly anarchic makeup. Besides, it's fun."

keeping the brain cycling
Finding the fun in tough challenges is another key to Wilkes' success, something that he extends into his pursuits outside his work.

He's recently qualified for his pilot's license and spends one night a week learning the art of pottery. He's also interested in Gothic architecture, hosting the Web site of a group of fellow enthusiasts at his HP web page.

"With Gothic architecture and art," Wilkes explains, "you can get quite intense about looking at things and understanding what's in front of you in a different way from technical things. Pottery is a very different non-mental concentration; I find it relaxing -- and frustrating in new ways, too. And the flying is this nice combination of intellectual and physical activity. A lot of brain cycles are required."

solving problems
Wilkes foresees plenty to keep his brain cycling in his future at HP Labs.

Clearly relishing the thought, he looks forward to being able to work with his team on "a never ending stream of exciting problems, problems whose value you can see. They're not just abstract problems, but ones where you can see how if this were solved, things would get better."

by Simon Firth