A Feature-Based Framework for Developing and Provisioning Customizable Web Services

A customizable web service is a service that enables service consumers to dynamically determine variants of the service they receive. Provisioning customizable services helps to efficiently address functional variability in customer requirements. However, this is challenging due to: i) the complexity in deriving the right subset of service capabilities for a service variant and ii) the existence of a large number of variants and their dependencies. We propose a feature-based framework to tackle this challenge. In our framework, a feature model is used to capture functional variability in customer requirements at a high-level of abstraction and to provide customers with a much simpler way to customize an atomic service. A service engineering process is designed to facilitate the systematic identification and implementation of variability during service development, and to maintain the mapping between variabilities at the feature modeling level and the service implementation level. We define a generative middleware that supports service deployment and exploits the mapping to enable runtime service customization. A large scale case study based on the Amazon web services is used for evaluation. In addition to addressing the challenge in provisioning customizable services, our experiments show that the generative middleware helps to reduce runtime resource consumption.