A PROBABILISTIC APPROACH TO EFFICIENTLY AND SUSTAINABLY LOAD BALANCING FOR CLOUD COMPUTING OPTIMIZATION

In order to lower environmental effect in cloud computing, sustainability calls for effective resource management. Often ignoring server heterogeneity, traditional load balancing techniques include the Power-of- d paradigm result in poor performance and significant energy consumption. This work presents an adaptive, energy-aware load balancing method dynamically distributing workloads depending on several server parameters: bandwidth, memory, latency, CPU speed, connection count, and cost. Task allocation is guided by a probabilistic model derived from Ant Colony Optimization (ACO). Furthermore, advanced metaheuristics like both Particle Swarm Optimization (PSO) and Grey Wolf Optimization (GWO) as well as conventional algorithms like Round Robin and Least Connections are greatly outperformed by the suggested method according to experimental data. It reduces response times by 18%, uses resources up to 22%, and runs 15% less energy. This method aligns performance goals with environmental responsibility and presents a sustainable and effective solution for cloud infrastructure.