Energy-aware load balancing in cloud environments using graph neural networks and grey wolf optimization.

Area of Science:

• Cloud Computing
• Artificial Intelligence
• Optimization Algorithms

Background:

• Cloud infrastructures require balancing resource utilization and energy consumption under dynamic workloads.
• Existing machine learning load balancing methods face challenges with training costs, generalizability, and real-time adaptation.

Purpose of the Study:

• To develop a hybrid framework for energy-aware and scalable cloud load balancing.
• To improve real-time adaptation to workload variations and enhance sustainability.

Main Methods:

• A hybrid framework combining Graph Neural Networks (GNNs) for workload representation and Grey Wolf Optimization (GWO) for task-to-VM mapping.
• GNN models intricate relationships between virtual machines, tasks, and resources.
• GWO optimizes mappings to minimize load imbalance, energy use, and task completion time.

Main Results:

• Reduced energy consumption by 18-27%.
• Decreased task completion time by 12-20%.
• Improved resource utilization balance by 15-22% compared to existing methods.

Conclusions:

• The proposed GNN-GWO hybrid model offers an effective solution for energy-aware and scalable cloud load balancing.
• The framework demonstrates superior performance in energy efficiency, task completion, and resource balancing under dynamic workloads.
• The hybrid approach provides a sustainable and adaptive solution for modern cloud environments.