Dynamic resource management heuristics for minimizing makespan while maintaining an acceptable level of robustness in an uncertain environment
Date
2006
Authors
Ye, Bin, author
Jayaseelan, Arun, author
Maciejewski, Anthony A., author
Siegel, H. J., author
Smith, Jay, author
Mehta, Ashish, author
IEEE Computer Society, publisher
Journal Title
Journal ISSN
Volume Title
Abstract
Heterogeneous parallel and distributed computing systems may operate in an environment where certain system performance features degrade due to unpredictable circumstances. Robustness can be defined as the degree to which a system can function correctly in the presence of parameter values different from those assumed. An important research problem in resource management is how to determine a resource allocation and scheduling of tasks to machines that optimizes a system performance feature while delivering acceptable level of robustness. Makespan (defined as the time required to complete all tasks in a resource allocation) is often the performance parameter that is optimized in such systems. This paper presents a robustness metric for dynamic resource allocations where task execution times are uncertain. The goal of this research is to develop heuristics capable of dynamically mapping tasks to machines such that makespan is minimized and a specified level of robustness is maintained. This research proposes, evaluates, and compares ten different dynamic heuristics for their ability to maintain or maximize the proposed dynamic robustness metric in an uncertain environment. In addition, the makespan results of the proposed heuristics are compared to a lower bound.
Description
Rights Access
Subject
robustness
dynamic mapping
resource allocation
makespan
resource management