Two-Step Ahead Prediction for Elastic Cloud Control System
No Thumbnail Available
Date
2020-02
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Addis Ababa University
Abstract
Elasticity property of cloud computing enables dynamic provisioning of computing resources
in order to match changes in application demand at run time. Cloud elasticity
solution acquires or releases cloud resources automatically using either proactive or reactive
mechanism. Hybrid of the two approaches has been proposed in previous researches
to take their advantages and mitigate their drawbacks; however, proposed approaches do
not address proactive model inaccuracy before resource deployment and after system load
behavior changes. In this research, threshold-based deviation between two-step ahead
recursive predictions and observed values is proposed for decision making before scaling
actions in combining proactive and reactive models. In addition, Stacked Long-Short Term
Memory Recurrent Neural Network has been implemented as predictive model that can
learn linear and nonlinear relationships of time series. The predictive model is retrained
automatically with new observations at run time to adapt system load changes. To evaluate
the proposed approach, experiments are conducted using CloudSim Plus with real
system CPU usage. Empirical analysis of experimental simulation revealed significance
of proposed solution in alleviating proactive model inaccuracy, deployment delay and
oscillation to improve utilization efficiency and application performance. Moreover, performance
of online retraining demonstrated that predictive model can learn new changes
at run time to enrich proactive mechanism.
Description
Keywords
Cloud Computing,, Cloud Elasticity, Recurrent Neural Network, Long-Short Term Memory