Article: Virtual machine workload prediction using deep learning Journal: International Journal of Cloud Computing (IJCC) 2024 Vol.13 No.6 pp.549 - 565 Abstract: This paper presents a novel approach to optimise resource allocation in virtualised systems, aiming to maximise performance and minimise operational expenses. Leveraging deep learning models, specifically long-short-term memory (LSTM) and bidirectional gated recurrent unit (bi-GRU), the method focuses on forecasting CPU load patterns in virtual machines (VMs). Accurate predictions are crucial for proactive resource management in dynamic cloud-based infrastructures. LSTM and bi-GRU excel in handling time series forecasting due to their ability to detect temporal connections in sequential data. Using pre-processed historical CPU load data, the models undergo training with hyperparameter adjustments to enhance performance. Experimental results demonstrate that the proposed models outperform others, achieving lower average root mean square error (RMSE) values (0.05636) and mean absolute error (MAE) values (0.03721). Comparative analysis with LSTM, GRU, bi-LSTM, bi-GRU, LSTM-GRU, and bi-LSTM-GRU confirms the high predictive capabilities of LSTM and bi-GRU, with the bidirectional architecture of bi-GRU enhancing accuracy by capturing connections between previous and upcoming time steps. Inderscience Publishers - linking academia, business and industry through research

Title: Virtual machine workload prediction using deep learning

Authors: C.S. Abhilash; Chaithra; Veena Garag; H. Priyanka

Addresses: Department of Computer Science and Engineering, PES University, Bengaluru, India ' Department of Computer Science and Engineering, PES University, Bengaluru, India ' Department of Computer Science and Engineering, PES University, Bengaluru, India ' Department of Computer Science and Engineering, PES University, Bengaluru, India

Abstract: This paper presents a novel approach to optimise resource allocation in virtualised systems, aiming to maximise performance and minimise operational expenses. Leveraging deep learning models, specifically long-short-term memory (LSTM) and bidirectional gated recurrent unit (bi-GRU), the method focuses on forecasting CPU load patterns in virtual machines (VMs). Accurate predictions are crucial for proactive resource management in dynamic cloud-based infrastructures. LSTM and bi-GRU excel in handling time series forecasting due to their ability to detect temporal connections in sequential data. Using pre-processed historical CPU load data, the models undergo training with hyperparameter adjustments to enhance performance. Experimental results demonstrate that the proposed models outperform others, achieving lower average root mean square error (RMSE) values (0.05636) and mean absolute error (MAE) values (0.03721). Comparative analysis with LSTM, GRU, bi-LSTM, bi-GRU, LSTM-GRU, and bi-LSTM-GRU confirms the high predictive capabilities of LSTM and bi-GRU, with the bidirectional architecture of bi-GRU enhancing accuracy by capturing connections between previous and upcoming time steps.

Keywords: virtual machines; VMs; long-short-term memory; LSTM; bi-GRU; CPU load prediction; cloud computing.

DOI: 10.1504/IJCC.2024.143524

International Journal of Cloud Computing, 2024 Vol.13 No.6, pp.549 - 565

Received: 25 Oct 2023
Accepted: 08 Jan 2024
Published online: 30 Dec 2024 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article

Title: Virtual machine workload prediction using deep learning

Keep up-to-date