By Steve Fisher, Editor In Chief -- Earlier this week the National Science Foundation announced a $12 Million middleware development project to create and deploy network services that will make it easier for Internet users to access a wide range of resources available through high-performance networks. NCSA will play a significant role in the project dubbed NMI (NSF Middleware Initiative) so Supercomputing Online interviewed Randy Butler, NCSA’s Division Director of Alliance Computational Environments and Security. Supercomputing: Please provide a brief description of the NMI initiative for the readers who may have missed the announcement. BUTLER: NMI is a project whose goal is to create advanced network services that are going to make things easier for a lot of researchers. It will allow them to access a huge variety of resources via the Internet and hopefully speed innovation in our industry…in our community. The effort will build on the successes achieved by the Globus project and the MACE initiative in developing middleware tools, and will integrate emerging middleware components into a comprehensive, commercial-quality, middleware distribution package that runs on multiple platforms. These middleware distributions will then be disseminated to research labs and universities worldwide. There are actually two groups who will receive the awards. One team, formed by Internet2, will include EDUCAUSE and the Southeastern Universities Research Association (SURA). The second team will consist of the University of Southern California School of Engineering's Information Sciences Institute (ISI), the University of Chicago, the National Center for Supercomputing Applications (NCSA), the University of California at San Diego (UCSD) and the University of Wisconsin at Madison. We will establish the GRIDS (Grids Research Integration Deployment and Support) Center. Supercomputing: How long has NMI been "in the works?" Where did the original proposal/idea originate? BUTLER: It would be hard to pin this down to a single proposal or idea. All of the organizations involved in this effort have years of experience in the development and deployment of middleware technologies. Efforts such as the Argonne National Laboratory/University of Chicago Globus project, University of Wisconsin's Condor, NCSA's Grid-in-a-box, Internet2's Glueworks, and many others have lead to this point. Supercomputing: Is the funding to be split equally by the Internet2 team and the GRIDS team? BUTLER: No. two-thirds of the funding is for the GRIDS Center and the other one-third is directed towards the Internet2 efforts. The GRIDS Center was actually two separate proposals that were linked together from the start. Carl Kessleman of the University of Southern California's Information Sciences Institute is the PI of one and I’m the lead for the other. Supercomputing: Please tell us the specific duties of the GRIDS center component of the initiative and the role NCSA will play. BUTLER: GRIDS Center's goals are: * To define an integrated, modular, and extensible Grid architecture that addresses a large fraction of current and projected middleware requirements for the scientific and engineering communities. To instantiate the NMI architecture, creating robust, tested, packaged, and documented middleware solutions that address a significant portion of these requirements, and that allow community extensions over the next several years to address other requirements. *To work with the middleware research community to evolve the architecture and to integrate selected components into our middleware infrastructure. *To provide production-quality software releases, support, training, and outreach, with the goals of ensuring that early-adopter user communities are successful and that new user communities are trained in the use of the infrastructure. *To develop tools and procedures to ensure that the resulting software is acceptable, deployable and supportable on a wide variety of end systems, including production environments on campuses and in laboratories. *To provide dedicated operations capability, providing 24x7 support for infrastructure elements, monitoring of grid infrastructure, and capture and analysis of grid usage data. In doing so, we enable the construction of a national-scale infrastructure that can be used by application communities such as NEES, GriPhyN, and others to explore the use of the middleware infrastructure on full scale, meaningful applications. NCSA will participate in the efforts to achieve all of these goals. It will lead efforts for the last three bullets listed above. ISI will lead efforts on the first three goals. Supercomputing: In the recent press release you're mentioned as comparing NMI and its expected impact to the original NSFnet. Would you care to expand on that statement? BUTLER: NSFnet physically connected together many US academic researchers, thereby establishing standards and consolidating effort. It allowed researchers to build and strengthen collaborations and to begin to think about virtual teaming. NSFnet spurred the interconnection of federal agency networks with each other and also with private networks. Standards, deployment and advocacy by many, many organizations, both federal and private, lead to the Internet as we know it today. The NSF Middleware Initiative will drive standards in an open fashion through the Global Grid Forum and IETF, integrate a useful set of capabilities into easy-to-deploy and support software distributions, and promote the use of these new capabilities through demonstrations with discipline specific research communities. NMI is nowhere near the physical scale of the original NSFnet, yet there is no doubt that Grids--which allow researchers to collaborate and share tools, resources, and data over networks--are a major way in which scientists will do their work in the years to come. There are currently Grids and grid software projects popping up all over the world, from CERN's Data Grid project in Europe, NASA's Information Power Grid, to NINF, a project in Japan that allows users to access all sorts of computational resources. Within the NCSA/Alliance community we have the In-a-Box software deployment efforts, which will bring software and hardware for Grid computing to even larger numbers, and we, together with many other partners are involved in the NEESgrid system integration project, which will create a virtual collaboratory for Earthquake engineers. There is also the TeraGrid project, funded by NSF's Distributed Terascale Facility award, the Alliance Virtual Machine Room, and our newest project the NMI. The effects will likely be much more subtle than the changes brought about by the Internet, but there is no question that Grids, and Grid-based research and collaboration, will have a major impact on the research community in the years to come. NMI is one of the projects that will help advance Grid efforts worldwide. It has already attracted the interest of, and cooperation from, other federal agencies and of researchers in Asia, Europe, and Canada. NSFnet gave the community a channel for communication, allowing researchers to share data and ideas. NMI will give researchers a common language to support advanced interactions such as multimedia collaboration, remote steering, and information discovery. ---------- Supercomputing Online wishes to thank Randy Butler for his time and insights. It would also like to thank NCSA’s Karen Green for her efforts in arranging this interview. ----------