An Efficient Resource Allocation Strategy for Virtualized Cloud Data Centres

Abstract

Virtualized cloud data centers host heterogeneous applications whose workloads change continuously, making static resource allocation inefficient and often responsible for server under-utilization, excessive energy use, performance degradation, and unnecessary virtual machine (VM) migrations. This paper proposes an adaptive resource allocation strategy that combines multi-dimensional utilization monitoring, dynamic host classification, SLA-aware VM selection, and energy-conscious VM placement. The strategy observes CPU, memory, bandwidth, and workload intensity, detects overloaded and underloaded physical machines, and reallocates VMs only when the predicted benefit exceeds the migration cost. A weighted allocation score is used to rank candidate hosts according to residual capacity, expected power increase, load balance, and SLA risk. The methodology is designed for evaluation in a CloudSim-based virtualized data-center environment using heterogeneous hosts and time-varying workloads. Performance is assessed through energy consumption, average utilization, SLA violation rate, number of migrations, response time, and active-host count. A representative simulation-style comparison indicates that the proposed strategy can improve utilization and reduce energy consumption and SLA violations compared with First Fit, Best Fit, and threshold-only consolidation. The study demonstrates that efficient cloud resource allocation should be treated as a multi-objective and adaptive decision problem rather than a single-resource bin-packing task.