A New Scheduling Algorithm for Servers

Abstract

Slowdown is used to measure the fairness degree of a scheduling algorithm in existing work. However, the fairness degree should be considered within a scheduling algorithm; rather than being tired with system environment. An innovative fairness model is proposed named hereby to intuitively measure the fairness of a server within through a scheduling algorithm. The new scheduling algorithm is called MPQ-LP which compromises both the fairness of PS and the highest performance of SRPT. Namely, MPQ-LP can easily adjust the performance and fairness degree of the scheduling algorithm. In addition, MPQ-LP allows each job have the fixed completion time with less computational complexity compared to SRPT when assigning the jobs to the queues. Simulation tests have been conducted to support our model verification