Two adaptive location policies for global scheduling algorithms

Abstract

Two important components of a global scheduling algorithm are its transfer policy and its location policy. While the transfer policy determines whether a task should be transferred, the location policy determines where it should be transferred. Based on their location policies, global scheduling algorithms can be broadly classified as receiverinitiated, sender-initiated, or symmetrically-initiated. Unfortunately, choosing from among these classes of algorithms is difficult, because how well each performs relative to the others and the effect each has on system stability has been shown to depend on the system load. In this paper, we present two new location policies that, by adapting to the system load, capture the advantages of all three classes of algorithms. A key feature of these location policies is that they are general, and can be used in conjunction with a broad range of existing transfer policies. Using simulation, two representative algorithms making use of these adaptive location policies are shown to be stable and to significantly improve performance relative to non-adaptive policies.

A Study of Load Distribution Algorithms in Distributed Scheduling

“…The most popular proposed concept for transfer policy is based on a optimum threshold. [1], [3], [4], [5]. Thresholds are nothing but units of load.…”

“…A widely used method by the researchers is polling, [1], [3], [4], [5], [6], which involves querying another node whether it can accept a task to be transferred to it. The variations are:  Random: It is a simple dynamic technique where a task is simply transferred to a random node.…”

Assignment schemes for replicated services in Jini

“…Following previous related work [5,12,14], one can identify two main branches of load balancing. The static load balancing (thoroughly covered in [5]) is suitable for clients arriving by group i.e.…”

“…underloaded and particularly idle nodes). Thus, load balancing recognises three types of balancing initiatives [14]:…”

“…Presumably, it is not the service's task to form complementary pairs and to swap the most delayed clients. At the same time nothing prevents the system from redistributing dynamically the clients already in service (or waiting for service in the local queues) according to algorithms in [3,12,14].…”