EWPTNN: An Efficient Workload Prediction Model in Cloud Computing Using Two-Stage Neural Networks
K Dinesh Kumar, E. Umamaheswari · Procedia Computer Science · 2019
Allocating insufficient resources to the cloud applications, leading to the loss of revenues, consumers and, Quality-of-Services (QoS). Another side, allocating resources more than is needed leading to the wastage of energy and cost to maintain the resources like servers, datacenters and, network bandwidth etc. So, this problem can be solved with predictive scaling methods by using machine learning approaches, which can estimate the future needed resources and, performing scaling operations in advance. To perform accurate scaling operations, authors presented an enhanced prediction model based on a neural networks, which is a combination of a classification and prediction model to predict the accurate utilization of resources. At the first stage, the proposed model categorizes the resources into three classes like over, normal and, under adaptively based on the utilization of resources. In the next stage, it predicts the future utilization of resources by using neural network regression model based on the classification results. The experimental results showed that the proposed model achieved accurate results.