Real-time IaaS Cost Model and Estimation Results for Cloud PC Laboratory Service

IaaS is well known as a very cost effective computing service which enables required infrastructures to be rented on demand without ownership of real hardware. It is very suitable for price sensitive services due to pay-per-use style. Operators of such services would want to adjust utilization policy promptly by estimating costs for cloud infrastructures as soon as possible. However, instantaneous response is not possible due to that cloud service providers offer at least several hours delayed billing information. Our work proposes a real-time IaaS cost estimation model based on resource usages monitored by virtual machine instance. We had operated PC laboratory service on a public cloud during full semester to validate our suggested model. From our experiments, averaged disparity between estimation and actual cost is less than 5.2%.