Evaluating services on the cloud using ontology QoS model

Abstract

To achieve large scale reuse and economy of scale, software are wrapped into services and migrated to the cloud platform. As a result, the quality of software services are affected by many third party components such as infrastructure services and platform services. In the collaborative framework, the stake-holders may have distinct, even contradictive, quality objectives. It is necessary to build agreements on quality properties, and to evaluate at runtime system's conformance to its service level agreements from different views. The paper proposes a QoS (Quality of Services) model for software services deployed on the cloud. The model identifies QoS concerns from three dimensions: system resource utilization, service performance, and price. Ontology language is used to describe these quality concerns and their properties. Based on the quality model, the paper proposes a method for service satisfaction evaluation from the end users' view point, taking all the three categories of factors into consideration. Experiments are conducted on Amazon EC2 platform using SPECWeb Support benchmark. It analyzes how each factor contributes to the end systems' satisfaction for different EC2 instance configurations.