If theres one person who may be able to unite the likes of Microsoft Corp., IBM, Sun Microsystems Inc. and other vendors fighting for control of the Web services space, its Ian Foster, the so-called guru of grid computing and the co-head of the Globus Project, a U.S. government- and industry-backed research initiative leading efforts to integrate grid technologies with Web services. Grids, and the work now under way at Globus, might hold the key to solving quality-of-service and availability problems that loom for Web services-style computing. As a result, vendors such as IBM and Microsoft are increasingly turning to the work Foster and his colleagues have done with grid technologies in the science and technical arenas.
eWeek Labs Senior Writer Anne Chen recently spoke with Foster about the future of commercial grids and the efforts being made to ensure grid technologies remain open standards. Besides leading Globus, Foster is associate division director, senior scientist and head of the Distributed Systems Lab in mathematics and computer science at Argonne National Laboratory, in Argonne, Ill.
eWeek: The idea of accessing computing resources and data without having to own computers isnt new. What has changed in the last few years to make the idea of grid technologies more viable to enterprises?
Foster: Certainly, people were talking about these general notions in the 60s at the University of California at Los Angeles and at a number of other places. A few things have changed in the last few years. First of all, broadband networks have certainly gotten things moving in the science space, along with the recognition that high-speed networks were qualitatively different things.
Then there is the ubiquitous availability of the Internet and, therefore, increased experience with … collaborative modes of working. And, most recently, people in business have been watching whats been going on in scientific and technical computing and have realized that there are some pretty interesting technologies applicable to the problems they have been trying to solve. That is the nature of the Globus Project, where various companies such as IBM, Sun and Microsoft are now participating. Theyre interested in seeing how they can exploit the technologies weve been developing and the open-source, open-standards community weve built to address problems they want solved.
eWeek: How will grid technologies change the way enterprise computing is done?
Foster: There seem to be multiple interests driving enterprises to delve into grid computing. One is the increased interest in a belief that its time to realize the notion of utility … computing. Enterprises are looking at the outsourcing of computing, storage and various services and hoping to use grid computing technologies as a way of realizing those capabilities.
There is also an increased interest in collaborative computing, in business-to-business computing scenarios and the recognition of fundamental resource sharing. Grids could be key enablers for powering all of these things. And then, particularly when we talk to people at IBM, there is this vision of autonomic computing. I would express it as computer systems getting more and more complex because people are trying to do increasingly complex things. This is really resulting in the decomposition of the traditional machine room model and the realization that one needs to start developing systems that are able to manage themselves to a greater extent. The technologies we in the grid community have been working on address some of these concerns in terms of providing enablers, for example, for self-describing mechanisms for service, and resource characteristics for resource management.
eWeek: You have spoken about the increasing integration of grid technologies with previously distinct commercial technologies such as Web services. What needs to be done to make this integration work? What other challenges do commercial grids face?
Foster: Were putting a lot of effort right now into integrating what weve learned with Web services within what we call the Open Grid Services Architecture. And now, companies like IBM and Sun and so forth are committed to integrating OGSA protocols into products like IBMs DB2 and Suns Sun Grid Engine. So ideally, if this is successful, we will achieve another step forward in terms of our ability to build large-scale, interoperable systems. One challenge is that in order to get broad adoption of these technologies, its not enough just to have IBM incorporate them. We need to get Sun to port them to [Java 2 Enterprise Edition], and BEA [Systems Inc.] and others to port their products as well.
Another set of challenges relates to how we ourselves in the Globus Project are committed to building an open-source code base for all this grid technology. We are committed to open standards because the grid is like the Internet. It is about the interoperability of systems. I dont think its feasible for anyone to develop a proprietary grid technology solution.
At the same time, we realize companies are going to want to incorporate these technologies into proprietary products and will want to differentiate themselves with the various quality-of-service levels. We need to find a way to allow them to do this without hindering interoperability.
For more information on Fosters grid work, go to the Globus Project at www.globus.org and the Global Grid Forum at www.gridforum.org.