EVALUATION OF PATTERN - ORIENTED REAL.-TIME EVENT SERVICE

Abstract

The CORBA Event Service provides a flexible model for asynchronous communication among objects. However, the standard CORBA Event Service specification lacks important features required by real-time applications. To resolve these limitations, Distributed Object Computing (DOC) group has developed a pattern-oriented Real-time Event Service as part of The ACE ORB (TAO) project [6]. TAO Real-time Event Service extends the CORBA Event Service specification to satisfy quality of service (QoS) needs of real-time applications. In this thesis challenges faced while designing TAO Real-Time Event Service are discussed. Performance of this service is evaluated and discussed. An optimization pattern is proposed and implemented to improve consumer-filtering scalability in TAO Real-time Event Service. This pattern specifies to construct sub-lists of consumers subscribed and search in it instead of searching the whole list when an event has to be forwarded. Experiments are conducted for evaluating the consumer-filtering scalability of TAO Real-time Event Service. The results produced demonstrate that it is possible to improve the scalability and hence performance by sacrificing memory foot print. The proposed optimization pattern is implemented and its performance is evaluated using TAO ORB on Windows XP with MS Visual C+ +. Rational Rose 98 is used for preparing the models in the UML notation.