Investigating Resource Allocation in a Standards-Based Grid Compute Economy

Published

October 31, 2007

Author(s)

Christopher E. Dabrowski

Abstract

This paper reports work investigating resource allocation in a simulated grid compute economy where consumers and providers employ software components implemented with emerging WS and grid standards. The investigation employs a model to simulate inter-actions between large numbers of consumers and providers where both parties dynamically enter into contracts to allocate resources. A series of trials is conducted to vary provider criteria for accepting contracts to compare performance using profit-based criteria to performance using more traditional criteria guided by system utilization under conditions of moderate load and overload. The results show that a grid computing economy, in which consumers and providers attempt to maximize individual monetary profit, can produce resource allocations using standards-based components that are not only efficient and benefit direct participants economically, but are also beneficial to the larger community that depends upon the computing grid. The results also show that the grid computing economy also performs well under overload conditions, while utilization-based resource selection strategies perform comparably well under moderate load while incurring reduced overhead. Measurement methods that employ time series analysis is used to support these conclusions by providing detailed insight into resource allocation over time.

Citation

- 7463

Report Number

7463

Pub Type

NIST Pubs

Keywords

distributed resource allocation, grid compute economy, grid computing standards, service level agreement, web services

Citation

Dabrowski, C. (2007), Investigating Resource Allocation in a Standards-Based Grid Compute Economy, - 7463, National Institute of Standards and Technology, Gaithersburg, MD (Accessed April 20, 2024)