Scalability refers to the extent of configuration modifications over which a system continues to be economically deployable. Until now, scalability of resource management systems (RMSs) has been examined implicitly by studying different performance measures of the RMS designs for different parameters. However, a framework is yet to be developed for quantitatively evaluating scalability to unambiguously examine the trade-offs among the different RMS designs. In this paper, we present a methodology to study scalability of RMSs based on overhead cost estimation. First, we present a performance model for a managed distributed system (e.g., Grid computing system) that separates the manager and managee. Second, based on the performance model we present a metric used to quantify the scalability of a RMS. Third, simulations are used to apply the proposed scalability metric to selected RMSs from the literature. The results show that the proposed metric is useful in quantifying the scalabilities of the RMSs.
A. Mitra et al., "Measuring Scalability of Resource Management Systems," Proceedings of the 19th IEEE International Parallel and Distributed Processing Symposium, Institute of Electrical and Electronics Engineers (IEEE), Jan 2005.
The definitive version is available at http://dx.doi.org/10.1109/IPDPS.2005.277
19th IEEE International Parallel and Distributed Processing Symposium
Electrical and Computer Engineering
Article - Conference proceedings
© 2005 Institute of Electrical and Electronics Engineers (IEEE), All rights reserved.