by December 5, 2002 0 comments

Since high-end applications run on network servers, it’s important to keep track of their consumption levels. So things like CPU utilization, amount of memory being used, etc must be measured. This information can be obtained from system log files and by knowing the behavior of frequently running applications.

Application monitoring is a key aspect of network auditing. It involves keeping a tab of performance levels of the mission critical applications running on your network, notably, application server, database server, web server, and higher end applications such as ERP. A system administrator must therefore monitor these applications to ensure that they don’t go
down. For this, certain tools are needed specifically for monitoring application performance. 

Applications can be monitored
from three different places, namely the user’s desktop, network and finally on the application server itself. In order to measure the performance, application monitoring software first construct a baseline, the regular usage pattern of server resources. Then, network and server performance metrics are measured in real time and compared to this baseline to get an idea of the usage at any given instant. 

The process of monitoring from the desktop/client machines is known as active monitoring. In this, each client machine runs an application agent, which gathers the statistics of all applications that are running and reporting it to a centralized server. While this gives a real-time picture of what’s happening on each client machine, it results in a lot of network traffic. Another way is passive monitoring, where the agent just sits on each client and monitors applications quietly. The moment an application’s behavior becomes abnormal, it sends the information to the server. 

The second place for application monitoring is in probing the network. This approach monitors both client and server traffic
and analyzes how they function. Unfortunately, this approach can be time consuming on a busy

The final place to do application monitoring is on the server itself. It takes traffic from the network, as well as from 
each client to measure the performance. 

KDE System Guard
Here we’ll see an application-monitoring software, called KDE System Guard that is based on a particular Linux command called “top”. This command lists the most CPU intensive processes running in the memory. It gives the option of killing the most resource hungry processes, indicates the zombie or idle processes in memory, and indicates CPU usage. The System Guard can be customized to show processes of a specific user, sequence the display according to age of processes, numerical order or resource consumption. Any application that is fired in the memory is immediately reflected. Moreover, it also indicates the utilization of swap space.

Being network enabled, it can monitor local and remote hosts. The system, and user processes can be viewed separately thereby making it more user-friendly. The software can be invoked from the K>System menu or by typing ksysguard in a terminal. The application consists of the ‘sensor browser’ and ‘workspace’. Registered hosts and their sensors, each of which monitors a system value, can be seen in a tree form on the sensor browser. You can have the option of this hierarchical in the workspace if need be. This feature enables the administrator to understand the hierarchy and correlation of the processes before performing some action on them. 

Varun Sharma

No Comments so far

Jump into a conversation

No Comments Yet!

You can be the one to start a conversation.