Workstations uptime analysis framework to identify opportunity for forming ad-hoc computer clusters

Abstract

Many solutions have been proposed to harness underutilized workstations to form ad-hoc and low-cost computer clusters. The success of forming ad-hoc computer clusters greatly depends on the usage characteristic of an organization. An organization with hundreds of workstations does not necessarily imply that it has the opportunity to form an ad-hoc computer cluster because different workstations may have different turned-on durations and these workstations may not be available at the same time depending on their users' usage. The objective of this work is to devise a framework to predict when and which workstations may be turned-on and for how long based on previously monitored uptime. 97 workstations have been selected from a real environment for 42 days to test the proposed framework. From the test results, it showed that there are indeed turned-on patterns on weekly basis. Some workstations have more consistent repetitive turned-on patterns while some do not, and this work created a framework to distinguish these workstations. The contribution of this work is the creation of a framework that allows administrators to identify potential workstations for forming ad-hoc computer clusters of their environment.