A Study on Load Balancing for Large-scale Distributed Computing Based on Improved Deep Learning

Some large-scale distributed computing load balancing methods suffer from excessive service violation rate. A large-scale distributed computing load balancing method based on improved deep learning is designed to improve the above problem. It takes the approach of offloading some of the downstream computing tasks, identifies large-scale distributed computing flow architectures, introduces the concept of linear functions, constructs a server energy consumption model, records the timestamps of state changes separately in sequential order, and designs load balancing methods based on improved deep learning. Experimental results: The mean service violation rates of the large-scale distributed computing load balancing method in the paper, and the other two large-scale distributed computing load balancing methods are: 16.016%, 19.487% and 18.688% respectively, indicating that the performance of the designed large-scale distributed computing load balancing method, after incorporating the improved deep learning algorithm, is more perfect.