A Novel Multi-Dimensional Resource Allocation Mechanismin Cloud
The cost of renting a medium level of cloud services from the Amazon Elastic Compute Cloud (EC2) web service is approximately $0.12/hr. Therefore, a resource-allocation strategy that efficiently allots cloud-server resources to clients involves increasing the resource use of each cloud server. This reduces the number of cloud servers rented by cloud-service providers, decreasing the rental costs of customers. Therefore, designing an efficient resource-allocation method is crucial. Resource-allocation problems are considered from one-dimensional or multi-dimensional perspectives. One-dimensional resource-allocation methods cannot compute the varying consumption levels of distinct resources such as the CPU, memory, and storage. Multi-dimensional resource-allocation methods have not addressed the consumption dependencies among various resources and allocating resources is considered an NP-hard problem. The proposed method improves the resource utilization of cloud servers up to 98% in multi-dimensional environment. The proposed method provides approximately 28.6% improved efficiency compared with that of the multi-attribute decision-based integrated resource-scheduling (MADIRS) method.
Index Terms- Cloud Computing, Resource Allocation, High Utilization.