A Hybrid CNN-LSTM Model for Virtual Machine Workload Forecasting in Cloud Data Center
Habte Lejebo Leka, Zhang Fengli, Ayantu Tesfaye Kenea, Abebe Tamrat Tegene, Peter Atandoh, Negalign Wake Hundera
Abstract
It is vital to precisely forecast the workload of Virtual Machines (VMs) to achieve efficient cloud resources management and reduce power consumption. In this research study, a deep learning-based hybrid strategy for VM workload prediction is proposed. To create an accurate prediction, the suggested prediction model integrated a convolutional neural network (CNN) architecture and a long-short-term memory (LSTM) neural network. The CNN component is used to elicit complex distinctive attributes of the VM workload data, while the LSTM component models temporal information and predict the future VM workload. Experimental results on real-world dataset have shown that the proposed CNN-LSTM model is effective on VM workload prediction, and when compared to frequently used workload prediction models, the proposed approach enhances workload forecasting performance.