A.o. Univ. Prof. Dr. Dipl.-Ing. eva Kühn
TU Wien

Richard Mordinyi

Managing Complex and Dynamic Software Systems with Space Based Computing

 

Dissertation, TU-Vienna, 2010

Abstract

Software systems are usually composed of distributed and heterogeneous application components representing higher-level business goals, and a middleware part abstracting the complexity concerns related to network and distribution.

 

In the course of developing complex software systems software developers have to deal with interacting application components and changing business requirements. The message-passing paradigm is a common concept allowing application components to interact with each other. But even asynchronous message-oriented middleware technologies are not suitable for complex coordination requirements since the processing and state of coordination have to be handled explicitly by the application component, thus increasing its complexity. Data-driven frameworks, like tuple spaces, support the coordination of application components, but have a limited number of coordination policies. Therefore, with respect to more complex coordination requirements application components still need to implement coordination functionality that is not directly supported by the coordination framework.

 

Middleware frameworks usually represent a specific architectural style and there is a dependency between application components and architectural style. In case a new business requirement demands the implementation of other architectural styles, the combinations of those styles further introduces additional cognitive complexity. Consequently, instead of a stable set of architectural concepts for effectively managing complexity concerns, the number of concepts a software developer has to work with explicitly increases with the size and degree of evolution of the system.

 

This thesis proposes the so called Space-Based Computing (SBC) paradigm to support software developers managing complexity issues regarding interaction requirements and agility of software architectures. The SBC paradigm defines an architectural style that flexibly combines and abstracts the properties of several architectural styles and extends them by sophisticated coordination models. In contrast to traditional coordination frameworks the approach supports exchangeable coordination models and regardless of the evolutionary degree or state of the system, it offers software developers a stable and limited number of architectural concepts to work with. The approach is evaluated in several industrial application domains: in air traffic management, production automation, and intelligent transportation regarding feasibility, effort, robustness, performance, scalability, and usability. The evaluation was carried out by means of prototype implementations, studies, benchmarks, and theoretical proofs. The results show a higher coordination efficiency, improved robustness against changing requirements, simplified realization of business requirements, and reduced complexity in applications.

Contact
Sitemap
Home
top | XHtml 1.0 strict | last update: Jun 2015