Design And Optimization Of Transformer-Enhanced Energy-Aware Load Balancing Techniques for Green Cloud Computing Environments

Abstract

Demand for computational resources has skyrocketed due to the ever-increasing popularity of cloud computing services, which in turn has raised energy consumption, operating expenses, and the negative effect on the environment. Efficient load balancing is now crucial to increase resource efficiency and cut down on energy consumption in modern cloud data centers. However, heuristic procedures or static scheduling rules are frequently employed in traditional scheduling and load balancing methods, which cannot meet the need of load balancing for frequent and unpredictable workload patterns. This restriction leads to an imbalance in resources, unnecessary virtual machines (VMs) migrations, increased power usage and degraded quality of service (QoS). A paradigm for sustainable cloud computing environments based on Transformer is introduced in this research, which is named Transformer-Based Multi-Objective Energy-Aware Load Balancing (TMO-EALB). It seeks to overcome these challenges. The proposed method can correctly anticipate the future workload conditions based on the historical and real-time data of cloud resources by recognizing the long-range temporal relationship using the Transformer network. A multi-objective optimization model based on the expected workload makes an intelligent scheduling decision. It reduces energy usage, virtual machine migration overheads, violations of Service Level Agreements (SLAs), increases resource utilization and maintains a balanced workload distribution. In order to provide adaptive resource allocation, the architecture constantly takes into account a number of cloud performance metrics, such as CPU utilization, memory use, network traffic, power consumption, and migration cost. To evaluate the proposed framework, we have performed experiments using CloudSim under various workload scenarios and observed their performance. The simulation results demonstrate the advantages of comparing the suggested method with the traditional load balancing methods when it comes to workload balancing, reducing unwanted migrations of VMs, reducing total energy consumption and improving the energy efficiency of the system. The proposed framework offers a sustainable and scalable solution for intelligent cloud resource management, making it easier than ever to manage resources.