DESPOT: Enhanced Dynamic Process Management for Beowulf Clusters on the Grid

Purpose

DESPOT is an NSF-funded CAREER project, award number EIA0092839.

Current systems for managing workload on clusters of workstations, particularly those available for Linux-based (Beowulf) clusters, are typically based on traditional process-based, coarse-grained parallel and distributed programming. In addition, most systems do not address the need for dynamic process migration based on differing phases of computation. The DESPOT project will first build a sophisticated thread-level resource-monitoring system for computational, storage and network resources. We will then use this information in an intelligent scheduling system to perform adaptive process/thread migration within the cluster. At the National Computational Grid level, we will integrate our system with metacomputing systems such as GLOBUS via a link to the NWS, in order to explore new scheduling heuristics based on the more detailed information provided by DESPOT.

People

• Principle Investigators
• Dr. Daniel Andresen, KSU
• Research Assistants
• Sadanand Kota
• Madhu Tera(to-do list)
• Tim Bower
• Ashish Sharma
• T.J. Rothwell
• Subbu Rangakrishnan
• Jeff Lebak
• Ethan Bowker
• Barrett Langton

Articles and presentations

• D. Andresen, D. Sexton, K. Devaram, V. Prasad Ranganath, ``LYE: a high-performance caching SOAP implementation,'' to appear in the Proceedings of the 2004 International Conference on Parallel Processing (ICPP-2004), Montreal, Quebec, Canada, August 15-18, 2004.
• D. Andresen, J. Lebak, E. Bowker, ``An Evaluation of Distributed Scheduling Algorithms Within The DESPOT Architecture,'' to appear in the Proceedings of the 2004 International Conference on Parallel and Distributed Processing Techniques and Applications (PDPTA'04), Las Vegas, NV, June 21-24, 2004. (PDF)
• K. Devaram, D. Andresen, ``SOAP Optimization via Parameterized Client-Side Caching,'' in the Proceedings of the IASTED International Conference on Parallel and Distributed Computing and Systems (PDCS 2003), pp. 785--790, Marina Del Rey, CA, November 3-5, 2003. (PDF)
• A. Hanenkamp, D. Andresen, ``Heterogeneous Channel Bonding Revisited,'' to appear in the Proceedings of the IASTED International Conference on Parallel and Distributed Computing and Systems (PDCS 2003), pp. 387--392. Nominated for IASTED Best Paper Award in the area of Communication Issues. Marina Del Rey, CA, November 3-5, 2003. (PDF)
• V. Ranganath, D. Andresen, ``A Set-based Approach to Packet Classification,'' in the Proceedings of the IASTED International Conference on Parallel and Distributed Computing and Systems (PDCS 2003), pp. 889-894. Nominated for IASTED Best Paper Award in the area of Networking. Marina Del Rey, CA, November 3-5, 2003. (PDF)
• K. Devaram and D. Andresen, SOAP Optimization via Client-Side Caching, in the Proceedings of the First International Conference on Web Services (ICWS'03),} pp. 520--524, Las Vegas, NV, June 24-27, 2003. (PDF)
• D. Andresen, N. Schopf, E. Bowker and T. Bower, Distop: A Low-overhead Cluster Monitoring System, in the Proceedings of the 2003 International Conference on Parallel and Distributed Processing Techniques and Applications (PDPTA'03),} pp. 1832--1836, Las Vegas, NV, June 24-27, 2003. (PDF)
• S. Prashanth, D. Andresen, Using implicit fitness functions for genetic algorithm-based agent s cheduling, in the \emph{Proceedings of the 2001 International Conference on Parallel and Distributed Processing Techniques and Applications (PDPTA'2001)}, pp. 445--450, Las Vegas, June 26-29, 2001. (PDF)

Links