Adaptive scheduling strategy for cloud computing resources based on Q-learning algorithm

Given the frequently varying loads of virtual machines and swift variations in resource demand in a cloud computing environment, this article introduces an adaptive resource scheduling strategy model based on Q-learning. The model develops a refined state and action space through real-time monitoring of virtual machine resource states, and an optimized reward mechanism that dynamically adjusts the resource allocation strategies. The reinforcement learning algorithm produces the optimal scheduling strategy through ongoing learning and adjustments to improve resource utilization, load balancing and task execution efficiency. Experimental results illustrate that the CPU utilization of virtual machines implementing the adaptive resource scheduling strategy using Q-learning achieves 0.87; memory utilization achieved 0.72; bandwidth utilization achieves 0.74; and the resource utilization curve is reasonably stable. This demonstrates that this model improves the efficiency of resource scheduling for virtual machines in a cloud computing environment, and ensures resources are efficiently and smoothly allocated.