| With the development of computer networks, it becomes more and more challenging for monitoring computer networks especially internal networks. Traditional network monitoring softwares always have fixed features and it’s difficult to customize.This paper proposes a network data analysis subsystem including macro network data analysis and microscopic network data analysis for data of application layer according to the features and drawbacks of existing network monitoring softwares. Macro network data analysis obtain data of SNMP from switches to analyze network failure and recovery event. Microscopic network data analysis bases on user-defined protocol description files to analyze data of application layer.The main content of this paper is to design and implement the two part of analysis in the subsystem. Macro network data analysis need to collect SNMP data with different level and adapt to virtual switching system. It maintains the history and current state of the switch. Then it does fault-analyzing by comparing to the default topology of the network based on the SNMP data and use the dynamic threshold algorithm and burst-detection algorithm. Finally, all data and events will be save in database for user to query. Microscopic network data analysis need to design and implement a description language for protocols of application layer. It supports character protocols and bit-level binary protocols. Users can describe dedicated protocols with it. The analysis system will serialize and deserialize the data of application layer based on these protocols’description files. Microscopic network data analysis can be ported across multiple systems. After requirements analysis and researching the key issues, it comes the designing and implementing of the subsystem. This paper designs and executes functional and performance testing for the subsystem, the results prove the subsystem meets the requirements. Finally, it states a summary of the whole paper. |
PDF Full Text Request