The Research And Implement Of Self-Adapted And Extensible Unix Information Collection System

Along with the increasing scale and complicacy of the network,, it's important to manage the network effectively and keep the running quality of the network. Mainframe management is an important part of IP network management, generally speaking, SNMP is the main method to collect mainframe information. But for UNIX mainframe, SNMP could not get the whole information we need, because some of them are not supported by SNMP.This paper puts forward a SNMP-independent UNIX mainframe collection system , and gives out the structure of this method to make it accord with the demands of flexible collecting, self-adapted collection, and also does some research on some key problems and gives us the solutions to them.1) Self-adapted plug-in Agent: high extensible and loading self-adapted Agent based on semantic analysis mechanism. This mechanism implemented performance data collection.2) Universal semantic analysis framework: Defining of universal analysis primitive symbol and operation regular which described the semantic analysis of management indicators . Designed universal analysis engine which based on source data submitted by agent deployed on mainframe to analyse primitive symbol and get management indicators' value.On the base of the research and the solution given of the key problems listed above, this paper gives out the Java\C\Shell implement of the scenario. This mechanism based system had already acted on a financial sector and got good results.