Dynamic multi-resource monitoring for predictive job scheduling.

Date of Award


Publication Type

Master Thesis

Degree Name



Computer Science


Computer Science.



Creative Commons License

Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License
This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License.


Standard job schedulers rely on either the user's estimation, or a few approaches that use performance databases to keep information about job runtimes to predict future runs. Co-scheduling for improved resource utilization, however, requires more detailed information as regards behavior on multiple resources to make predictions about slowdowns. Thus, information about communication, I/O, and computation at application level is needed but hard to estimate by the user. Furthermore, dynamic adaptive resource allocation requires information about the different processes on different machine nodes. We present an intelligent monitoring tool, ScoPro, which provides such information. To make monitoring more feasible, ScoPro harnesses the dynamic instrument techniques, which postpone insertion of instrumentation code until the application is executing. To keep intrusion low, we limit monitoring to short test phases. (Abstract shortened by UMI.)Dept. of Computer Science. Paper copy at Leddy Library: Theses & Major Papers - Basement, West Bldg. / Call Number: Thesis2005 .L586. Source: Masters Abstracts International, Volume: 44-03, page: 1407. Thesis (M.Sc.)--University of Windsor (Canada), 2005.