What kind of applications do you expect to be driven by this work? If you're an enterprise, and you're trying to
set up a system that will use resources in multiple parts of a company to support workload, you need to be able to dynamically connect the computing in [one] place with data in another [place]. Perhaps this is allocated to computing to another part of your company. You need to be able to provision this all to protect against failures and provide fail-over, and that's where OGSA comes in. It's about addressing issues of how you discover properties of services in your network, or how do you monitor the status of these capabilities. You have to negotiate service level agreements with various components of the organization, etc. What companies have been most important to the development of grid computing to this point? IBM is putting in the most effort to this point, but certainly HP, Oracle and Sun are all very engaged. Then there is a variety of smaller companies like Platform and Entropia or United Devices. Microsoft is still in the watch-and-see mode, but I think they get it. Where is the intersection of grid computing and Web services? The open grid services architectural (OGSA) effort is working to take grid technologies and integrate them into a Web services framework. In a sense, that work is integrating management capabilities into Web services. The concept of Web services already speaks to invoking potential in a distributed environment, but it doesn't address issues of manageability, and that's what OGSA is doing. I think it's a very synergistic relationship. How do you think grid computing will impact the enterprise applications market -- efforts such as ERP or CRM? It should enable new solutions and business models based on outsourcing and de-emphasizing the current coupling between providers of computing and their users. Initially you should see, within enterprises, more and more de-coupling of production and consumption of computing, and increasingly people will work out how to build various forms of service provider offerings. No one has been able to do that effectively to this point. How far have users come in terms of tapping into the potential for grid computing? In the sciences, people have been applying grid computing for some time. In industry, it's mostly intra-enterprise solutions from companies like Platform, IBM and Sun. This work is mostly around resource management or making more efficient use of computing resources, and also for data federation.
We're starting to see more innovative large scale deployments like Butterfly.net, a company building platforms for online video gaming. The idea is to use large computing resources -- provided by companies like IBM -- as a service provider to run large populations of online video games. Why do you think that is?
First of all, the technologies aren't in place to allow the real large economies of scale to be exploited. Secondly, the technologies for managing systems with components that are geographically distributed haven't been put into place. The grid computing community is addressing both of those issues right now. How long before we view grid computing as a widespread business strategy?
Well, how widespread is 'widespread?' By the end of this year, people like IBM will be rolling out success stories. A lot of it is experimental, but I think it's going to be going commercial by 2004. There are already some resource management tools out there that provide basic capabilities. More sophisticated commercial technology will arrive this year, and by next year I think you'll see more significant deployment.
FOR MORE INFORMATION:
Article: Gartner says IM hot, grid computing getting warmer
Tip: Finding utilitarian uses for utility computing What advice would you give CIOs in approaching the concept of grid computing on a practical level?
There are companies like Charles Schwab that are already experimenting with grid computing internally. It's not too early to begin looking at areas where you can make more efficient use of existing resources or enabling new capabilities by coupling resources. Starting internally on a small level is the most promising direction.