A Deep Recurrent Network for Web Server Performance Prediction

Abstract

Recurrent neural network(RNN) has been widely applied to many sequential tagging tasks such as natural language process(NLP), and it has been proved that RNN works well in those areas. In this paper, we propose to use RNN with long short-term memory(LSTM) units for web server performance prediction. Classical methods focus on building relation between performance and time domain, which can't capture the essence of web server performance. In this paper, we analyze the log of nginx web servers which contains user's url access sequence, and predict the performance of the servers by using RNN-LSTM. Experiment result shows that our model gets a good performance in predicting web server performance on the data set which has been deployed in online service.