Performance-Aware Hybrid Throttled-ESCE Load Balancing with Two-Stage Load Threshold Mechanism for Heterogeneous Cloud Environments

Load balancing (LB) is a critical determinant of performance and efficiency in cloud computing environments. While various algorithms like Throttled Load Balancing (TLB) and Equally Spread Current Execution (ESCE) exist, they suffer from inherent limitations such as inefficient resource utilization, a lack of capacity-awareness, and the risk of Virtual Machine (VM) overload, leading to suboptimal performance. This paper introduces a novel hybrid LB method that synergistically combines the core principles of the TLB and the ESCE algorithms. The proposed method employs a sophisticated two-stage load threshold mechanism to intelligently distribute incoming tasks. It systematically prioritizes VMs by first selecting from idle pools based on processing power, then from partially loaded VMs using a capacity-weighted score, and finally from a broader pool based on historical throughput efficiency, all while strictly enforcing a load threshold to prevent VM overload. By explicitly accounting for heterogeneous VM capacities (MIPS, CPU cores, RAM) and real-time efficiency metrics, the method ensures a more balanced and intelligent workload distribution. Comprehensive simulations conducted using the CloudAnalyst simulator demonstrate the superior efficacy of proposed approach. The results show a significant 44.97% average improvement in Data Center (DC) processing time and a 23.38% reduction in average response time to clients compared to traditional algorithms (ESCE and TLB). These findings conclusively establish that the proposed method effectively mitigates the limitations of its predecessors and offers a robust, high-performance solution for enhancing QoS in modern heterogeneous cloud DCs.