GRIDS Center and CDIGS
Ian Foster, PI
Goals: Globus development began in 1996 as an effort to provide the scientific community with a set of tools and services for building distributed applications. Thanks to the recent NSF award for the Community Driven Improvement of Globus Software (CDIGS), developers from the University of Chicago (led by Ian Foster) and University of Southern California (led by Carl Kesselman) are able to continue contributing to Globus development. CDIGS project goals include: evolving and enhancing Globus functionality, performance, scalability and robustness; improving usability and manageability so as to decrease the cost and complexity of deploying, operating and using Globus infrastructure; supporting major NSF users and communities; expanding the Globus community.
Significance: Developed over the last 10 years, the Globus Toolkit software has emerged as a key tool for many NSF science and engineering projects working to realize such new approaches. Globus security, resource discovery, resource access, and data management functions underpin major NSF projects in mathematics and physical sciences, biological sciences, geosciences, and engineering. The software fills many roles, providing essential infrastructure functions in projects such as TeraGrid, Open Science Grid, and the Network for Earthquake Engineering Simulation; serving as a development platform for applications such as distributed simulation, data analysis, and collaboration; and acting as an invisible end-user tool for many scientists and engineers. Globus software is also used extensively in computer science research and education, and supports many international scientific collaborations. A growing U.S. and international community of Cyberinfrastructure users, developers, and deplorers now depend on Globus software. The “Community Driven Improvement of Globus Software” (CDIGS) project is intended to meet the needs of this community for continued improvements to that software.
Accomplishments: CDIGS is primarily a development project, not a research project. It is designed to produce, enhance, and support software that will enable scientific advances by other groups, in both discipline sciences and computer science. Nevertheless, the proposed work will inevitably result in not only better software but also new knowledge. As project participants work with users to understand requirements, integrate known techniques into software, develop new techniques where existing techniques are inadequate, and struggle with practical issues that arise during software construction, testing, and deployment, they will inevitably gain new understanding of technical problems and identify both new problems and new solutions to problems. The value of this project derives above all from the large size and considerable ambition of the Globus user community. Dozens of major NSF science projects use Globus software, either directly or via higher-level tools or packages that incorporate Globus components. Many other faculty and students use Globus software via campus or national infrastructures. Major international collaborations gain benefit from the common use made of Globus software. Many university courses are taught using Globus software, and many graduate students build on Globus software in their research. All users also have requirements that current Globus software cannot meet, for example for greater functionality. CDIGS provides the resources needed to meet these requirements, and thus enhance the scientific productivity of those users. Furthermore, by providing a context for rational long-term planning, CDIGS enables projects with long planning horizons to make sensible decisions about software adoption.
