A Bisection Scheduling Algorithm Using Run-time Prediction Based on Grid Computing

摘要

Grid computing is a major research area with increasing impact on the scientific and business field. Since a grid connects large-scale geographically distributed computers, scheduling in such environment is a complex undertaking. In order to simply the scheduling work, we propose a resource filter policy to choose appropriate resource set for the task firstly. Then we present a bisection scheduling policy that uses predicted mean and variance resource capacity information to make task-mapping decisions. This can not only simplify the large-scale resource scheduling issues but also reduce the task execution time