A Methodology for Dynamic Scheduling of Divisible Workloads in Grid Environments

Scheduling divisible workloads in distributed systems has been one of the interesting research problems over the last few years. Most of the scheduling algorithms previously introduced are based on the master-worker paradigm. However, the majority of these algorithms assume that workers are dedicated machines, which is a wrong assumption in distributed environments such as Grids. In this work, we propose a dynamic scheduling methodology that takes into account the two prominent aspects of Grids: heterogeneity and dynamicity. The premise of our methodology is to use a prediction strategy to estimate the CPU speed of each Grid resource and subsequently feed this estimation to a static scheduling algorithm in order to divide the workload into suitable chunks in light of the available computational power. Such integration can produce dynamic scheduling algorithms that can handle the constantly changing properties of Grid resources. Key-Words: Grid Computing, Divisible Workload, Dynamic Scheduling Algorithm, CPU Speed Prediction.