The National Computational Science Alliance (NCSA) has awarded OSC $6,000 to fund an undergraduate researcher. The researcher will develop utilities and applications to support Access Grid user services.
The Partnership for Advanced Computational Infrastructure – Research Experiences for Undergraduates (PACI REU) Program is designed to encourage interactions among students and scientific mentors working on common projects. The program hopes to engage a diversified pool of talented undergraduate computational science research students with PACI partners. Its goal is to develop the next generation of computational scientists.
“This program is a wonderful opportunity to expose talented undergraduates to opportunities in innovative research,” said Scott Lathrop, NCSA Education, Outreach, and Training Division Associate Director.
The PACI REU grant will allow OSC to take the next step in developing tools outlined in the Global Grid Forum (GGF). GGF is a group that is developing documentation on grid technology specifications and best practices.
“Establishing grid computing environments changes the user services staff’s primary role. Additional tools and analysis techniques will be required to effectively support the grid user community,” said Jim Giuliani, OSC Systems Engineer and Project Lead for GGF’s Grid User Services Research Group. “Through our work with the Global Grid Forum, OSC is able to take an active role in defining the grid user support system structure.”
The purpose of PACI REU is to develop additional utilities to support users in a production Grid environment. High performance computing centers have support staff members in place to aid users with problems developing applications on one of their platforms. Most support staff members have an assortment of debugging and analysis software tools at their disposal to assist in diagnosing the causes of various problems. As computational Grids move from research projects to production systems, support staff members must have the proper software tools to diagnose user problems. While some of the existing software tools can be applied to Grid environments, additional tools will be needed.
This project seeks to develop several user support software tools for Grid environments. By evaluating these tools on existing applications, this project will also help outline the additional support tools needed to support Grid systems.
Project goals are:
- To educate students in the fundamentals of Grid computing and the structure of Grid middleware, most specifically the Globus Toolkit.
- To understand the execution environment of Grid applications and where information is logged from each Grid service used.
- To develop a software tool that will collect all log entries for a Grid application. A method for organizing and browsing this information will also be developed.
- To develop a software tool that gives a snapshot of all resources allocated to a job.
- To develop an interface allowing support staff to interrogate specific resources dynamically and probe specifi Grid components.
- To evaluate the new software tools and determine which user services needs are met and if additional tools are needed.
Project goals follow the recommendations of the “Support Services and Tools Requirements” document being developed by Giuliani. The research student will contribute to the technical content of this document and be invited to attend the eighth GGF meeting in June, when this document will be reviewed. This project will focus on the Programming Tools & Environments area. Collaboration with the Interaction Environments area is encouraged, since possible tools may be incorporated into user environments to allow users to self-diagnose problems. Tools developed as part of this project will be tested on the Alliance Grid Testbed (AGT). OSC is a partner in the Alliance and is active in the AGT. During development, tools will also be available to Alliance TeraGrid (ATG) members and national TeraGrid production systems for evaluation and comments. AGT team members include: the Ohio Supercomputer Center, National Center for Supercomputing Applications (NCSA), Boston University, University of Kentucky, University of Wisconsin, Argonne National Laboratory, and the High Performance Computing, Education and Research Center (HPCERC) at the University of New Mexico.
More information on PACI REU is at http://alliance.osc.edu/reu/ or http://www.npaci.edu/Outreach/PACI-REU/.
