Adaptive Virtual Instance Consolidation for Enhanced load Balancing in Cloud Data Centers

Cloud computing offers significant benefits to commercial clients by efficiently handling expanding workloads in a planned manner. The primary enabling technology for cloud computing is virtualization, which simplifies infrastructure management and usage. This paper leverages virtualization to promote green computing and allocate resources based on workload requirements. By minimizing skewness and mixing diverse workloads, server utilization is increased. However, managing client demand poses challenges, which are addressed through Virtual Machine (VM) technology to dynamically allocate resources. The implementation of a virtualized environment is expected to reduce job response times and optimize task execution based on resource availability. VMs are assigned to users according to their needs, and VM live migration facilitates VM and Physical Machine (PM) mapping. Efficient utilization of cloud resources helps balance workloads and prevent system slowdowns. A local negotiation-based VM consolidation technique is employed to anticipate task requests and generate virtual space. A co-location strategy combines small, empty rooms to enhance server performance, while a self-destruction strategy eliminates inaccurate data based on time-to-live property. Real-time resource allocation and an online prediction model aid in determining partition sizes for reduction jobs and dynamically allocating resources to reducers with large partitions to expedite task completion.