Human network experts and managers try to resolve these problems, but because networks are large, complex, and both geographically and functionally distributed, a lot of time may be required to resolve the problems. Even if a manager finds the symptom of a problem, the cause may be located in another, remote network, or the manager may not be able to obtain enough information for troubleshooting. Moreover, the manager must often investigate a huge amount of flowing packet data and must often cooperate with other network managers to isolate the real cause of the problem. To do this, however, may take a long time. In internetworks, some problems sometimes show no apparent symptoms (these types of problems are called potential problems. Potential problems will probably cause serious effects when the network becomes larger or when more hosts are attached to it. Furthermore, a few problems may spread to other network segments through IP routers or gateways.
There are a number of SNMP/MIB-based network management systems. These are useful to understand the network state for expert network managers. However, this type of systems do not show when a problem occurred, and what is the problem. We also take into account that some companies do not have the experts where such a SNMP-based management system do not work as intended.
These discussions confirm the need for a knowledge-based system capable of diagnosing the problems in internetworks.
Back to the LODES top page