Co-organizer: Karl M. Göschka, Schahram Dustdar, Mehdi Jazayeri, Vienna University of Technology, Austria
Abstract:
Large scale systems often significatly lack of scalability of their non-functional properties like performance or dependability. Given a particular set of technologies, these properties can usually be traded against each other to optimize some of the properties while relaxing others. However, the question remains, how to reach an overall system optimum by choosing an apropriate architectural composition of strongly-coupled data-centric system parts (middleware like CORBA, EJB, COM+) and loosely-coupled service-centric system parts (e.g. P2P, GRID, Web services). Certainly, simplicity is a key issue for scalability, while complexity seems to be the outcome of current system engineering practice. Therefore, one way to tackle the inherent complexity of large-scale systems is to maintain the initial simplicity and robustness of loosely coupled systems, while at the same time guaranteeing those quality properties needed by real-life applications (e.g. performance and dependability) by coordination between services. Our workshop consequently focuses on hybrid large-scale systems and how to increase fault-tolerance, reliability and availablity of such systems.
