The current role of network management, which can be seen mainly in the support of isolating faults and detecting performance bottlenecks by long term load monitoring, has to be extended when using ATM technology. By taking into account the possibility to control the traffic in the network, the aim of network management should be to prevent faults, i.e., cell losses, and to support efficient resource allocation in the ATM network. So - in opposite to traditional network management issues - this work takes a step further by stating that the aim of monitoring a given set of parameters (e.g. load on a link, discarded cells on a connection) cannot only be to recognize the violation of QoS requirements, but also to detect potential quality decreases. With such an approach it should be possible to act in time - instead of reacting after problems appeared - and thereby prevent the occurrence of performance bottlenecks, quality decreases and even error conditions in the ATM network.
To reach this objective a very detailed structuring and interpretation of management information is needed. This means, parameters to monitor and actions to be carried out when certain values of the parameters are exceeded have to be selected very carefully, as the aim of avoiding cell losses and supporting optimal resource allocation should not lead to sending an uncontrolled number of alarms to the management station.
In this work a model is presented, which allows to derive the management concepts needed here, based on local circumstances. Therefore the required structuring of management information is derived from concrete management goals, which can be obtained when using a 'connection-oriented view' of the network. Afterwards, it is outlined how the critical values of the parameters can be chosen based on the (dynamic) traffic characteristics in the network. Finally, critical aspects about the proposed approach are discussed, as for example the necessary small interval for the collection of traffic information.
A suitable environment to further refine and approve the above approach is given by another just started DFN-funded project, the German RTB-Project (Re- gional Testbed) where complex ATM networks are used in production environments. Results will be pointed out together with first implementation experiences on a SNMP-based platform.
Summarizing, this work tries to outline how it could be possible to pull out network management from just 'monitoring' equipment behavior to active support of network quality and efficient resource allocation in future broadband networks, closely integrating practical experiences.