Using Machine Learning Algorithms for Cloud Client Prediction Models in a Web VM Resource Provisioning Environment

Abstract

In order to meet Service Level Agreement (SLA) requirements, efficient scaling of Virtual Machine (VM) resources in cloud computing needs to be provisioned ahead due to the instantiation time required by the VM. One way to do this is by predicting future resource demands. The existing research on VM resource provisioning are either reactive in their approach or use only non-business level metrics. In this research, a Cloud client prediction model for TPC-W benchmark web application is developed and evaluated using three machine learning techniques: Support Vector Regression (SVR), Neural Networks (NN) and Linear Regression (LR). Business level metrics for Response Time and Throughput are included in the prediction model with the aim of providing cloud clients with a more robust scaling decision choice. Results and analysis from the experiments carried out on Amazon Elastic Compute Cloud (EC2) show that Support Vector Regression provides the best prediction model for random-like workload traffic pattern.

A Comprehensive Analysis of Accuracies of Machine Learning Algorithms for Network Intrusion Detection

“…Supervised learning deduces a functional relationship from training data that generalizes well to the whole dataset. In contrast, unsupervised learning has no training dataset and the goal is to discover relationships between samples or reveal the latent variables behind the observations [11]. Semi-supervised learning falls between supervised and unsupervised learning by utilizing both labeled and unlabeled data during the training phase [10].…”

“…Semi-supervised learning falls between supervised and unsupervised learning by utilizing both labeled and unlabeled data during the training phase [10]. Among the three categories of machine learning, supervised learning is the best fit to solve the prediction problem in the auto-scaling area [11]. Therefore, this research focuses on supervised learning.…”

“…One of the original authors of UNSW-NB15 was also involved in creating the BoT-IoT dataset [11], as depicted in Fig. 2, in the Cyber Range Lab of UNSW Canberra Cyber Center.…”

“…The training dataset has 2,934,817 records. The features used for the experiments were the top 10 features [11] selected by the creators of this dataset.…”

Multi-formalism Models for Performance Engineering

“…With respect to paradigms other than those reviewed in this paper, machine learning algorithms have achieved a considerable popularity to predict the performance of cloud architectures. In [24], the authors discussed issues related to scaling of VMs resources in cloud computing implementing proreactive strategies based on neural networks, linear regression, and support vector regression, the latter providing the best accuracy.…”