Adaptive, Efficient and Fair Resource Allocation in Cloud Datacenters leveraging Weighted A3C Deep Reinforcement Learning

Cloud data centres demand adaptive, efficient, and fair resource allocation techniques due to heterogeneous workloads with varying priorities. However, most existing approaches struggle to cope with dynamic traffic patterns, often resulting in suboptimal fairness, increased latency, and higher energy consumption. To overcome these limitations, we propose a novel method called Weighted Actor-Critic Deep Reinforcement Learning (WA3C). Unlike static rule-based schedulers, WA3C continuously learns from the environment, making it resilient to changing workload patterns and system dynamics. Furthermore, the algorithm incorporates a multi-objective reward structure that balances trade-offs among latency, throughput, energy consumption, and fairness. This adaptability makes WA3C well-suited for modern multi-tenant cloud infrastructures, where diverse applications often compete for limited resources. WA3C also supports online learning, allowing it to adapt in real time to shifting workload compositions without the need for retraining from scratch. The model's architecture is designed to be lightweight and scalable, ensuring feasibility even in large-scale deployments. Additionally, WA3C introduces a priority-aware advantage estimator that better captures the urgency of tasks, enhancing scheduling precision. As a result, WA3C achieves more effective convergence, lower latency, and balanced resource allocation among jobs. Extensive experiments using synthetic job traces demonstrate that WA3C consistently outperforms both traditional and reinforcement learning-based baselines, highlighting its potential for real-world deployment in large-scale cloud systems.