Combined Load Prediction Model of SVM-LSTM based on Markov Chain Error Correction

Abstract

Generally, load balancing is implemented based on static load balancing strategy or dynamic load balancing strategy. Static load balancing strategy lacks real-time performance, while dynamic load balancing strategy can select the appropriate server to process requests according to the real-time performance of the server. With the development of machine learning and deep learning, models can be built to predict the load of the server in the future, and the performance of the server can be known in advance. In this paper, a combined load prediction model of SVM-LSTM based on Markov chain error correction is proposed by combining support vector machine and LSTM. Experiments on public data sets show that the experimental results of this model are better than the single prediction model and the combined prediction model without Markov error correction.