ANFIS with natural language processing and gray relational analysis based cloud computing framework for real time energy efficient resource allocation

Abstract As one of the most commonly used types of application, command line applications have aroused extensive attention. Predicting command line application workload and providing reasonable resource scheduling strategy can effectively improve resource utilization and performance of VMs. This paper provided a novel scheme with two-level hybrid adaptive model to predict VMs load based on the strong regularity of command line applications, and elastically configured the CPU and memory resource for VMs. Instead of time series prediction, we extracted and analyzed the feature attributes of command line applications by natural language processing (NLP) technology, and used the gray relational analysis (GRA) to implement attribute reductions. Then a two-level hybrid adaptive model was designed to predict the VM load efficiently and accurately, including CPU and memory load. Bayesian algorithm was used for classification to select the applications that increase the CPU of a VM by more than 5%, then giving the CPU and memory load prediction of VMs by aANFIS model. Extensive experiments demonstrated that the elastic allocation strategy can improve the VMs performance and resource utilization.