University of California, Irvine. ICS 243D – Internet Technologies - Spring 2001

Roberto Silveira Silva Filho ID# 85849631

The Network monitoring tool is composed by a set of daemons (operating system level processes) that collect management information from the hosts in a distributed system, and a concentrator server, that periodically collect this information in the monitored servers. In order to be monitored, each host (server) of the system executes a data collection daemon. Some examples of this information collected include: CPU usage, users logged in the machine, free memory, network connection status and so on.

The data is collected locally using unix scripts and commands as dmesg (shows the startup log of the system), vmstat (shows the cpu load), finger (shows the users logged in), netstat (shows network statistics) and so on. Once executed, the output of these scripts are processed in order to collect, classify and store the performance data in a local database.

Periodically, the concentrator agent polls the network administration daemons and collects the performance information. This data is processed in order to extract statistic and temporal information about the network.

An administration client, using a browser, can query the concentrator and collect the administration information that is presented, using graphics and tables, in a human readable way.

The architecture of the system is depicted below.

The daemons and concentrator serves will be implemented in perl. The information will be stored in plain text formats in the local hosts and in the concentrator server. The user interface will use a web browser.