Cloud computing has emerged as a popular paradigm for provisioning on-demand and scalable computing resources. However, efficiently managing the allocation of Virtual Machines (VMs) to tasks and handling overload/underload situations remains a key challenge. This paper proposes a new task migration strategy to enable efficient resource utilization in cloud data centers. The strategy uses lightweight virtualization technology to reduce migration overheads and an optimization model based on queuing theory and control theory for making informed migration decisions. The experiments on real-world Google cluster workload traces demonstrate that our strategy can significantly improve resource utilization while meeting service level objectives compared to state-of-the-art solutions. The proposed strategy consolidates tasks faster to idle servers through predictive analysis helping improve resource utilization and the results are compared based on Time (s) and Output Voltage (V) and Inductor Current (A).