Optimal Resource Allocation in Grid Environments

Grid development involves the efficient management of heterogeneous, geographically distributed, and dynamically available resources. In this environment, selecting resources and scheduling jobs in such a way that the application requirements are met, in terms of throughput and execution time, is a critical issue. Any resource allocation strategy requires the support of a suitable performance model for both the grid infrastructure and the application. The efficacy of the allocation policy depends on the quality of the underlying model. In this chapter we describe how queuing network models can be used to find optimal resource allocation for data parallel applications on hierarchical grid architectures. The proposed models rely on the statistical pattern of computation, communication, and I/O operations in the parallel applications, as well as on measurable infrastructure characteristics. We finally show how the high variability in the execution and communication times must be considered when modeling applications on geographically distributed grid infrastructures.