Job Scheduling Strategies for Parallel Processing

This volume contains the papers presented at the 10th Anniversary Workshop on Job Scheduling Strategies for Parallel Processing. The workshop was held in New York City, on June 13, 2004, at Columbia University, in conjunction with the SIGMETRICS 2004 conference. Although it is a workshop, the papers were conference-reviewed, with the full versions being read and evaluated by at least five and usually seven members of the Program Committee. We refer to it as a workshop because of the very fast turnaround time, the intimate nature of the actual presentations, and the ability of the authors to revise their papers after getting feedback from workshop attendees. On the other hand, it was actually a conference in that the papers were accepted solely on their merits as decided upon by the Program Committee. We would like to thank the Program Committee members, Su-Hui Chiang, Walfredo Cirne, Allen Downey, Eitan Frachtenberg, Wolfgang Gentzsch, Allan Gottlieb, Moe Jette, Richard Lagerstrom, Virginia Lo, Reagan Moore, Bill Nitzberg, Mark Squillante, and John Towns, for an excellent job. Thanks are also due to the authors for their submissions, presentations, and final revisions for this volume. Finally, we would like to thank the MIT Computer Science and Artificial Intelligence Laboratory (CSAIL), The Hebrew University, and Columbia University for the use of their facilities in the preparation of the workshop and these proceedings.

Inhalt

Parallel Job Scheduling - A Status Report.- Scheduling on the Top 50 Machines.- Parallel Computer Workload Modeling with Markov Chains.- Enhancements to the Decision Process of the Self-Tuning dynP Scheduler.- Reconfigurable Gang Scheduling Algorithm.- Time-Critical Scheduling on a Well Utilised HPC System at ECMWF Using Loadleveler with Resource Reservation.- Inferring the Topology and Traffic Load of Parallel Programs Running in a Virtual Machine Environment.- Multi-toroidal Interconnects: Using Additional Communication Links to Improve Utilization of Parallel Computers.- Costs and Benefits of Load Sharing in the Computational Grid.- Workload Characteristics of a Multi-cluster Supercomputer.- A Dynamic Co-allocation Service in Multicluster Systems.- Exploiting Replication and Data Reuse to Efficiently Schedule Data-Intensive Applications on Grids.- Performance Implications of Failures in Large-Scale Cluster Scheduling.- Are User Runtime Estimates Inherently Inaccurate?.- Improving Speedup and Response Times by Replicating Parallel Programs on a SNOW.- LOMARC - Lookahead Matchmaking for Multi-resource Coscheduling.