Raising network fault management intelligence

Most large network management centers have relatively low skilled personnel as their first level operations staff. Many organizations attempt to cope with this situation by restricting the set of problems these people have to deal with to those which are well understood and documented. Several software packages exist which can correlate and filter incoming events from the network and present a select subset to the operator. Unfortunately, programming these fault management applications requires considerable expertise and effort. Often, once the initial development is done, the implementation remains static, while the network itself is dynamic. This paper proposes a methodology for documenting known faults and responses, programming fault correlation engines, continuously examining real behavior, and feeding the result back into the programming process. This results in a continuous improvement in fault management intelligence, with corresponding improvement in network availability and thus value of the network to the organization.