Predicting Response Times of Applications in Virtualized Environments

We propose simple queueing models for predicting response times of applications executed in a cloud computing platform under the SaaS model. We assume that each application instance is executed within a virtual machine running on a computing node of a data-center, and that VMs running concurrently on the same node share fairly its capacity. Our main contribution is to explicitly take into account the different behaviors of the different classes of applications (interactive, CPU-intensive or permanent applications). We show that simple expressions of the mean processing times of applications can be obtained using standard results from queueing theory. Experiments on a real virtualized platform show that the mathematical models allow to predict response times accurately.