Betreuer: Hartmut Schmeck, Lei Liu
Forschungsgruppe: Effiziente Algorithmen
Beginn: 01. September 2009
Abgabe: 28. Februar 2010
Discovery in Service-oriented Computing Service-oriented Computing (SOC) provides fundamental means for integrating legacy, existing, and future technical IT systems into distributed enterprise applications. A full adoption of the design paradigm service-orientation may result in a loosely coupled, flexible, and interoperable enterprise IT architecture with high reusability. However, service-orientation does not provide by itself means to cope with the complexity incorporated in such an enterprise IT architecture. Various approaches have been proposed to enable self-organization in SOC to deal with the increasing complexity for human participants in the system. One of these approaches is to use the concepts of Organic Computing. A critical issue of enabling an Organic SOC (OSOC) is to have robust, reliable, and scalable algorithms for the discovery of essential components in large-scale SOC. Existing centralized discovery mechanisms such as UDDI may experience difficulties in large-scale service-networks. Typical problems like a performance bottleneck or a single-point-of-failure are the major points preventing an active adoption of such centralized mechanisms. To address these problems, different approaches have been proposed, such as in the WS-Discovery specification. The main aspect of these approaches is to move away from centralized systems.
Goal of the Thesis The goal of this work is to design and implement a decentralized approach to enable automatic distributed service discovery. Hence, one has to analyze 1) the requirements concerning discovery in largescale self-organizing SOC, 2) existing research conducted in the field (e.g. WSDiscovery), and 3) to design a reliable distributed discovery mechanism with respect to the self-organizing capabilities in OSOC. Furthermore, the designed discovery algorithm should be realized with a proof-of-concept implementation in an existing reference SOC simulation environment to prove its plausibility and capabilities. The resulting implementation should be able to keep the complete system landscape in a distributed and selforganized manner. The implementation should be carried out with up-to-date Java technologies, among other things, JDK 6 and eclipse 3.x. Furthermore, you have the possibility to work actively within an ongoing research project. If desired, you may work at the computer lab in our institute. The working languages in this thesis can be either German or English.