Intelligent filtering in network management systems

Abstract

Network management systems have to handle a huge volume of notifications reporting unprompted on events in the network. Filters that reduce this information flood on a per-notification basis fail to perform adequate information preprocessing required by management application software or human operators. Our concept of intelligent filtering allows for a highly flexible correlation of several notifications: Secondary notifications can be suppressed or a number of notifications can be aggregated. An intelligent filter was implemented using a rulebased language and was applied within SDH network management. Several modules, configurable while the filter is operating, support the user considerably and with excellent runtime performance. Further development is envisaged that provides for smooth integration into management application software.