PerformanceGuard collects data using PerformanceGuard agents on individual Windows computers—usually end-user computers. The agents collect response times and other performance metrics on these computers. Each agent assembles collected data into reports. At predefined time intervals the agents send collections of reports to the PerformanceGuard frontend server.
The frontend server unpacks the agents' reports and stores them in the frontend server database and/or hands the data to the backend server service that then inserts the data into the PerformanceGuard backend server database.
The backend server performs required aggregation of data and through the PerformanceGuard web server it makes data available for you to view in the PerformanceGuard web interface. The backend server also provides notifications and reports (click thumbnail to view image in full size):
Communication between agents and the frontend server takes place using TCP/IP. The frontend server listens on a single TCP port (default is 4001), and it's then the responsibility of the agents to contact the frontend server.
The frontend server never contacts an agent, and the agents don't listen on any public ports. If there are firewalls between agents and the frontend server, these should be set up to forward requests to the frontend server TCP port.
The frontend server and the backend database are connected using JDBC.
You can only have one backend server, but you can have multiple frontend servers if load distribution is required. In that case each frontend server must have its own database.
Timing Considerations
In order to be able to configure agents properly, it's important to understand the timing of an agent.
The agent collects performance data in reports. A single report describes the performance for a period of time, for example 60 seconds.
At predefined time intervals the agent sends reports to the frontend server. This is typically done every few minutes (the default is three minutes).
| Info | ||
|---|---|---|
| ||
| Example: If reports each cover 60 seconds, and reports are sent to the frontend server every three minutes, the agent sends three reports to the frontend server each time the agent connects to the frontend server. |
In order to collect the local performance metrics (CPU usage, memory usage, etc.), the agent samples the values at regular intervals, typically every one or two seconds.
| Info | ||
|---|---|---|
| ||
| Example: If the local performance metrics are sampled every other second, and reports cover 60 seconds, the average value for CPU usage is the average of 30 measurements, and the maximum value for CPU usage is the highest among the 30 sampled values. |
| Panel | ||||||
|---|---|---|---|---|---|---|
Search this documentation
On this page
In this section
|