A top request from most customers that I talk with is how to use EdgeSight to monitor the XenApp/Presentation Server environment. One thing that needs to be remembered about EdgeSight is that it monitors all applications that are launched on the device that is being monitored. This in turn includes the core applications that keep a XA/PS infrastructure up and running efficiently.

Here is a link to a fantanstic article on setting up Alerts in EdgeSight to monitor the necessary core apps.
EdgeSight Alerting for Citrix Presentation Server – CTX116449 - http://support.citrix.com/article/ctx116449

Below is a snippet taken from the document.

Infrastructure Health

There are many components that contribute to the overall health of a Presentation Server farm. By creating real-time alerts for the following processes as detailed below, an administrator can quickly determine the health of a particular server or the entire farm.

What to Monitor

Process Name Process Description
imasrv.exe Citrix IMA Service
xte.exe Citrix XTE Service
hcaservice.exe Citrix Health Monitoring and Recovery
termsrv.exe Terminal Server Service
smaservice.exe Citrix SMA Service
imaadvancesrv.exe Citrix Services Manager

How to Monitor

There are several recommended alert rules that can be created for each process, all of which are defined in the alert rules setup as Application Alerts. Each alert informs the administrator of a particular problem.

High application resource usage

After setting the CPU time (percent) or Memory usage (Kbytes) values for the particular process, EdgeSight will fire an alert when it detects that the process has exceeded those configured values. Processes consuming a high amount of CPU or memory could impact the overall performance of the server and have a negative impact on the end user experience.

Process fault

This alert will notify administrators when the process configured in the alert rule has faulted due to an unhandled exception within the process. The details collected by EdgeSight give a total picture of what was occurring on the server when the fault occurred without expending time and resources trying to reproduce the problem.

Thrashing application

By configuring the thread queue length value for a process in this alert, administrators can quickly be notified if there are any processor bottlenecks that may be degrading server performance. Thread queue length is invaluable in determining applications that are bottlenecked on a system as the parameter indicates the number of CPU work items currently waiting for the threads of the application to process them.