I propose to explore the design pattern of service-oriented architectures (SOA) especially concerning their maintenance and SOA related technologies. In this context I will develop a data loss prevention (DLP) system with a web service proxy for the Enterprise Measurement Infrastructure (EMI) with the necessary functions to improve maintenance tasks.
Every modern enterprise nowadays relies on software and software systems to collect and process data. Due to the highly embedding of those software into the working processes of each employee it is of great importance to maintain these system to prevent data loss and endangerment of their work. Furthermore maintenance in SOA is even more important as each service evolves independently and there can be big didifference in performance, availability and reliability. To ensure this there are a wide range of practices beginning with conventional design principles of object-oriented programming to the facilitation of
redundancy and/or scalability up to complex analysis of message pattern in the communication of the services.
The target of our concrete example of EMI was \to build better, easily extendable, and maintainable measurement systems which are easily integrated and aligned with modern business needs\. In order to achieve this EMI services are using gateways with a \push forward\ adaption which will cause a data loss in case of maintenance or absence of a gateway. This means we have to be prepared for planned failure cases like malformed data but also for unplanned cases like the failure of the database or even the failure of the whole system.
This thesis will design and development a robust system which recognizes the absence of gateways of EMI and prevent data loss through data caching with the help of a web service proxy. To further improve the maintainability we will provide a tool to automatically generate a description of the REST-based gateways for an easier deployment of the web service proxy. Moreover the design of the proxy will be extensible for new optional features (e.g. filter or automatic testing).
 Matthias Vianden, Andreas Steens, Horst Lichter. Towards a Maintainable Federalist
Enterprise Measurement Infrastructure. Proceedings of Conference of the 23nd International
Workshop on Software Measurement (IWSM), 2013.