Since both the number and the state of managed resources dynamically change (e.g. new objects are created, calls are established or dropped) it is necessary to constantly update the topology data that is contained in the management platform database. The traditional approach to this problem is the polling of the network for new resources and for state changes of already known resources by the management platform. The disadvantages of this approach are on the one hand the considerable delay until new resources are identified and, on the other hand, the generation of potential excessive network traffic that reduces the available bandwith for other applications. To avoid these problems we developed an approach that is completely event-driven. Managed resources emit events to well-known event channels when their state changes or when they notice the creation or deletion of other resources. To realize the generation of topology events we modified the following SOMObjects runtime components:
The benefit of this event-driven approach is that the management related network traffic is greatly reduced because an event is only generated when a change in topology data has really occurred and not proactively by the management platform. This is particularly useful for LEO/MEO satellites where changes occur frequently and where bandwith is limited and therefore expensive.