Best Practices for Monitoring Client Apps

paulbr

I've been tasked with creating some application monitors for our "critical" client apps such as Citrix, Email, and SharePoint.  I've looked over the many templates and done a bunch of searches but there seem to be several ways to go about monitoring such as WMI counters, Windows services or forms login.  I've run into some issues such as WMI not returning useful data during an outage and Windows services getting hung but still running so don't get any alerts.  I was wondering if anyone has run into similar issues and found what monitors seem to work the best.

I was also wondering what people do for creating an account to run component monitors as far as permissions and other security concerns.  I've been using the forms login monitor with my personal account but don't want to do that when going live with the monitors.  I was thinking of creating a "dummy" account that has read-only permissions to the apps but would have to find a way to lock down access to sensitive data.  Also, for monitoring OWA do you need a mailbox attached to the account?

Thanks for any feedback!

aLTeReGo

I'll let the members of the Thwack community chime in with their recommendations of what works best for them in their environment. I'll try and field the technical how/to.

I've run into some issues such as WMI not returning useful data during an outage and Windows services getting hung but still running so don't get any alerts.  I was wondering if anyone has run into similar issues and found what monitors seem to work the best.

Generally a combination approach is used when monitoring any application. Citrix is a complex distributed application that has many moving parts. As such, no single component monitor type will provide all the visibility required to ensure the application is both running and performing optimally. I recommend monitoring all critical services, processes, and relevant performance counters. I would then complement these monitors with TCP port monitors for application specific ports, and user experience monitors wherever possible to synthetically mimic user interaction with the application. These methods combined make up a fairly comprehensive application monitoring strategy.

I was also wondering what people do for creating an account to run component monitors as far as permissions and other security concerns.  I've been using the forms login monitor with my personal account but don't want to do that when going live with the monitors.  I was thinking of creating a "dummy" account that has read-only permissions to the apps but would have to find a way to lock down access to sensitive data.  Also, for monitoring OWA do you need a mailbox attached to the account?

Most customers choose to create service accounts for monitoring applications with SAM. These tend to be local or domain level administrator accounts that have all necessary permissions to monitor the specific host or application. If you need to use a non-administrator account you can follow the steps outlined in this KB here.