Cloud Resource Scheduling Using Semantic Search Engine Based on Improved Parallel Genetic Algorithm

Cloud resource has the features of dynamic, heterogeneous, distributed and complexity etc. Hence how to effectively schedule Cloud resources becomes a research focus in Cloud. Meanwhile the numbers of resources and tasks to be scheduled in Cloud are usually variable. This makes the Cloud resource scheduling a complex optimization problem. None of existing Cloud systems is both being an automated scheduling and considering the optimal usage of resources. To address these problems, we propose a Cloud resource scheduling strategy using improved parallel genetic algorithm (IPGA)-based Semantic Search Engine (SSE). SSE is a new type of service search engine developed by Semantic Computing laboratory in University of California, Irvine. It provides Cloud users with a friendly problem-driven interface to automatically schedule resources that would be used to build a solution according to users' requirements. Further we adopt IPGA in SSE to optimize the scheduling so as to obtain the optimal usage of resources. In our proposed IPGA there should be a code distant between the selected parents to retain the population diversity. Moreover, individuals can periodically migrate from one subpopulation to another. The architecture of our proposed scheduling strategy is presented as well as the process and implementation of IPGA. The experiment results show our proposed approach can reduce about twenty-one percent average tasks execution time due to the usage of SSE and IPGA can reduce about twenty-six percent average algorithm execution time and obtain better fitness with the generation increase comparing to the traditional genetic algorithm (TGA).