Mapping Parallel Applications with the Approach of Reducing Communication Overhead in Cloud Computing

Abstract

Implementation of parallel programs, divided into some tasks, and selection of proper resource among the available resources to execute these tasks are expressed as an important issue. This research aimed to evaluate a mapping method of tasks to resources, which are distributed in the cloud-computing environment. In addition, the tasks are categorized and their mapping is performed on the most suitable resource in terms of calculation and communication costs with regard to the communications between the tasks. The proposed method is compared to two Min-Min-C and Max-Min-C algorithms. Results of the performed scenarios indicate the reduction in the time required for mapping.