Priority Queue Scheduling Approach for Resource Allocation in Cloud

摘要

Resource utilization in cloud is very challenging because of its dynamic nature especially in heterogeneous applications. Even though, Virtual Machine (VM) technology permits multiple workloads to be processed simultaneously, it does not ensure application performance at certain scenarios. Therefore, cloud datacenter providers either do not render any performance assurance or choose static rather than dynamic VM allocation that results in ineffective usage of resources. This study tackles the problem of resource allocation inside a datacenter which runs various kinds of application workloads, specifically non-interactive and transactional applications. It specifically focuses on how available amount of resources such as memory and Virtual machine Central Processing Unit (VCPU) of the current cloud infrastructure have been utilized according to the user requests, projects and applications assigned to the cloud environment. The available resources are allocated to each VM both within and across VCPU. In this research, Priority Queue (PQ) scheduling algorithm is proposed. Fair share policies are defined at each queue to deal with dynamic priority of the requests submitted by the user. According to the dynamic priority of user requests, they are scheduled at two levels on the basis of their resource accessibility. The proposed scheduling algorithm hosts the virtual machines on cloud nodes to utilize the resources in a well-organized manner and the performance is evaluated and compared with conventional scheduling methods.