Our goal was to open the NetView Event Management Services (EMS) for CORBA events in addition to SNMP traps in order to create a single point of reception and processing of events from network, systems and application management. Our approach is based on the CORBA Event Service, which has been described in section III-C.2. We have decided to use event channel objects and typed event communication because we believe that the generic event format is insufficient for passing the complex information needed for managing CORBA-based distributed applications. Event channels have the following useful properties: The amount of event consumers and their object references are completely transparent to the suppliers; these send their events directly to an event channel and thus only need to know one reference (the one of the event channel) although an arbitrary number of consumers may receive the event. Typed event communication yields not only the advantage of being able to distinguish between application, network and resource events but presents also the opportunity of structuring the events stemming from CORBA agents according to its severity and w.r.t. different categories like fault, topology or security. This gives us the possibility to select events according to their kind and severity: it is feasible to use the push-model between the event channel and the consumer (i.e., the platform) for important events and to rely on pull-type communication for less critical events.
Two interfaces were defined between the CORBA environment and the management platform (see also figure 3):
We will discuss the implementation of both interfaces in the following sections.