{Runtime Variability in Online Software Products: A Comparison of Four Patterns
Jaap Kabbedijk, Slinger Jansen, Tomas Salfischberger · 2014
Abstract—Business software is increasingly moving towards the cloud. Because of this, variability of software in order to fit requirements of specific customers becomes more complex. This can no longer be done by directly modifying the application for each client, because of the fact that a single application serves multiple customers in the Software-as-a-Service paradigm. A new set of software patterns and approaches are required to design software that supports runtime variability. This paper presents two patterns to solve the problem of dynamically adapting functionality of an online software product; the Component Interceptor Pattern and the Event Distribution Pattern. Addi-tionally, it presents two patterns to dynamically extent the data model; the Datasource Router Pattern and the Custom Property Object Pattern. The patterns originate from case studies of current software systems and are reviewed by domain experts. An evaluation of the patterns is performed in terms of security, per-formance, scalability, maintainability and implementation effort, leading to the conclusion that the Component Interceptor Pattern and Custom Property Object Pattern are best suited for small projects, making the Event Distribution Pattern and Datasource Router Pattern best for large projects. Keywords—architectural patterns, quality attributes, software architecture, variability. I.