Optimized Weighted Round Robin Approach for Load Balancing in Cloud Computing

摘要

Technological advances in distributed computing paved way for unprecedented possibilities. Towards this end, cloud computing has emerged and capable of rendering computing resources on demand. In cloud computing it is inevitable to ensure that consumers' Service Level Agreements (SLAs) are met. To achieve this cloud computing exploits two specific concepts like load balancing and scheduling. As part of load balancing some Virtual Machines (VMs) are migrated for effective utilization of resources. The tasks scheduled may be of two types namely pre-emptive and non-pre-emptive. Nonpreemptive tasks are irrevocable unless they are finished properly. Therefore it is very essential to assign such tasks to VMs carefully in the first attempt itself. There are possibilities that in a VM multiple tasks, both independent and dependent, may run and thus create problems with nonpreemptive tasks. As the jobs arrive randomly and dynamically with heterogeneous loads, it is important to allocate them to resources in a better way. Since scheduling and load balancing are complementary in nature, in this paper, we propose a framework with underlying algorithms that take care of load balancing and optimized performance. Optimized weighted round robin is the approach employed for load balancing. We built a prototype application on top of CloudSim to simulate and demonstrate proof of the concept. The results revealed that the proposed methodology for load balancing is effective in the presence of nonpreemptive dependent tasks.