Composable cost estimation and monitoring for computational applications in cloud computing environments

With the offer from cloud computing providers, scientists have the opportunity to utilize pay-as-you-go resources together with their own and shared resources. However, scientists need to decide which parts of their applications should be executed in cloud computing systems in order to balance the trade-off between cost, time and resource requirements. In this paper, we present a service for estimating, monitoring and analyzing costs associated with scientific applications in the cloud. Cost models associated with different application execution models are proposed and these cost models can be composed to determine costs of different scenarios. We present techniques to estimate costs for service dependency and to monitor costs associated with typical scientific applications. Experiments with real-world applications are performed to illustrate the usefulness of our techniques. Our service could eventually be integrated into cloud resource management and execution services to support on-the-fly resource scheduling.