Optimization of Anti-Affinity Virtual Machine Placement Using Improved Game Theory Algorithm

In modern data centers, efficient virtual machine (VM) placement is essential for optimizing resource utilization and minimizing energy consumption. However, anti-affinity Constraints pose significant challenges. Traditional methods, such as First Fit Decreasing (FFD), often fail to effectively balance these constraints with energy efficiency. To address this, we propose GGAPA, a hybrid algorithm integrating game theory and genetic algorithms for optimized VM placement under antiaffinity constraints. In GGAPA, VMs are modeled as players in a strategic game, where game theory analyzes their interactions to guide placement decisions, while genetic algorithms evolve solutions to minimize energy usage while adhering to antiaffinity rules. We evaluate GGAPA across small (100 VMs, 50 PMs), medium (200 VMs, 100 PMs), and large (400 VMs, 200 PMs) data center scales. Experimental results show that GGAPA significantly outperforms FFD, standard genetic algorithms, and Grey Wolf Optimizer (GWO) in reducing energy consumption, demonstrating its scalability, robustness, and suitability for real-world data center applications.