Title: PPMaaS: performance-based pricing model as a service

Authors: Monika Simjanoska; Sasko Ristov; Marjan Gusev; Goran Velkoski

Addresses: Faculty of Computer Science and Engineering, Ss. Cyril and Methodius University, Skopje, Macedonia ' Faculty of Computer Science and Engineering, Ss. Cyril and Methodius University, Skopje, Macedonia ' Faculty of Computer Science and Engineering, Ss. Cyril and Methodius University, Skopje, Macedonia ' Innovation LLC, Skopje, Macedonia

Abstract: The pricing models currently established by the cloud service providers charge the consumers for renting virtual machines in a pay-as-you-use manner. The quality of service of the acquired resources is negotiated via service level agreements; however, the performance is almost never discussed. Cloud service providers aim to assign the available resources to as many consumers as possible, whereas the consumers intend to maximise the resources' utilisation. The two mentioned goals are often unpredictable and a good prediction model is of high concern. In this research we propose a new performance-based pricing model realised as a service (PPMaaS), targeting both the cloud service providers and the consumers. The decisive model is proposed as a result of machine learning analysis of various performance measurements obtained from different experiments done in both single and multi-tenant cloud settings. It is able to suggest the type of virtual machine with lowest CPU utilisation achieving the required service quality for a given load. The consumers will use the model to predict the best and cheapest virtual machine type for a given load and within requested response time limits.

Keywords: cloud computing; machine learning; pricing models; performance-based pricing; cloud service providers; virtual machines; performance measurement; quality of service; QoS.

DOI: 10.1504/IJRIS.2015.070908

International Journal of Reasoning-based Intelligent Systems, 2015 Vol.7 No.1/2, pp.16 - 25

Published online: 31 Jul 2015 *

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