Motivated by observation of the shortcomings of the existing work and the emerging new requirements, in particular from the application domains such as telecommunications, we have designed COBEA for event handling to extend the existing Cambridge Event Paradigm and the CORBA event service. The main goal is to design an architecture which provides a framework for object-oriented design and development of active application systems, especially in a large distributed environment. COBEA extends the CORBA Event Service, namely by supporting the publish-register-notify mode, parameterised filtering, fault-tolerance, access control, and composite events; all these features are missing from the CORBA event service. COBEA is a reincarnation of the Cambridge Event Paradigm with all the features mentioned above as well as support for dynamic addition of new event types and event mediator.
The basic requirement on an architecture for event handling is that events can be identified, classified, detected, specified and asynchronously reported to any interested party through a standard or application-defined interface. A general architecture, based on which large-scale distributed active application systems can easily be constructed, is clearly required. We believe that both direct and indirect event communication should be supported based on a wide range of application requirements, e.g. in the areas of CSCW (Computer Supported Cooperative Work), management in network, telecommunication and distributed systems, multimedia systems and mobile systems [2, 3, 6, 9, 13]. We focus on supporting the push model, i.e. after explicit registration of interest by consumers, events are pushed directly or indirectly by suppliers to consumers (a.k.a. the Notification Model). Event registration is essential for receiving selective event notifications. Such a scheme not only solves the main problems with synchronous communications - the saturation of network resources caused by polling operations, but also solves the problem of end user saturation caused by pushing everything through. The pull model can easily be supported by using existing technologies thus specific support is not necessary. The goals of COBEA are, in outline, as follows: