Implementation of Log Analysis System for Desktop Grids and Its Application to Resource Group-Based Task Scheduling

Abstract

It is important that desktop grids should be aggressively deal with the dynamic properties arisen from the volatility and heterogeneity of resources. Therefore, it is required that task scheduling should positively consider the execution behavior that is characterized by an individual resource. In this paper, we implement a log analysis system which can analyze the execution behavior by utilizing actual log data of desktop grid systems. To verify the log analysis system, we conducted simulations and showed that the resource group-based task scheduling, based on the analysis of the execution behavior, offers faster turnaround time than the existing one even if few resources are used.