QAComPS: A Quality-aware Federated Computational Semantic Web Service for Computational Modellers

This research sets out to help computational modellers to select the most cost effective Cloud service provider. This is when they opt to use cloud computing in preference to in-house HPC facilities. Cloud computing is a pay-per-use model for accessing computing resources from a variety of service providers such as Amazon EC2. Increasingly cloud providers are offering the high performance computing options that are necessary for computational modellers. The paper is concerned with a quality-aware computational broker (QABroker). The QABroker service federates across specific service packages offered by a selected set of computational cloud providers that potentially meet the user’s computational resource and QoS requirements. These vary during the various stages of the computational modelling cycle. The core of the QABroker is a novel Quality-aware federated computational semantic Web service (QAComPS). This includes an integrated ontologybased system that makes use of OWL2 features. This is used to filter the cloud providers’ services into three groups. These are: High, Medium and Low quality of service. This classification is then used by a MatchMaker to automatically select the highest ranked service that meets the user requirements. A SAWSDL interface was used to transfer semantic annotations to/from the QAComPS service and QABroker. Early evaluation of the QAComPS service was very promising and demonstrates its potential to make cloud computing more accessible and cost effective for computational modellers.