Parallel stress estimation for consistent task scheduling using buddy strategy

Abstract

Stable task assignment among processors is a shrewd effort because such approaches require priori information about current processor load and then perform allocation. Initial policy structure includes no information about processor load but as the task groups arrive and get scheduled, this information acts as a prerequisite for next allocation. The task of load stability is tricky when static scheduling strategies are followed in allocation because these strategies require supplementary efforts to balance the load as the approaches designed so far do not follow load stability criteria during task allocation; rather they follow task adjustment to balance the load later on that enriches the overhead of context switching of tasks among processors. Context switching efforts not only includes the efforts for changing process address space, but position in ready queue is also affected. The work projected in this desertion performs load stability with priori information about processor utilization, based on this metric value processor space is partitioned among different categories. Depending on the load status, processors are categorized and an appropriate set out of those is identified that act as buddy for others and handles incoming queue. In other words stress free “buddy set” handles incoming job queue in place of their “stress underneath buddy set” processors so as to eliminate the pains of context switching and tasks adjustments after the tasks allocation.