International Journal of Advances in Computer Science and Its Applications
Author(s) : LAKSHMIPRIYA.C , S. SATHYALAKSHMI
Grid has evolved as heterogeneous resource sharing environment. The resources are from different management domain. Finding a suitable resource for a given job is called scheduling. The highly dynamic and heterogeneous nature of grid makes scheduling a big challenge. To achieve maximum resource utilization and minimum makespan grid needs effective and efficient scheduling algorithm. Grid scheduling normally considers the availability of the grid resources and application demands. These demands are not known in advance for the highly dependent application and require more data transfer during execution. Without proper consideration of these demands at the time of scheduling leads to poor system performance and increase in makespan of the application. This unpredictable data transfer is called Uncertainty in communication. IPDT-FUZZY uses some parameters to find the uncertainty in given applications. When the degree of uncertainty is less, the IPDT-FUZZY scheduler behaviour is similar to the classical scheduler. In this paper Support Vector Machine (SVM) based scheduler is proposed to consider the uncertainty in communication demand, and also to compare its performance with other classical schedulers.