Table of contents
Monitoring Exchange 2007 Client Access Server role-specific counters
Disk performance countersAlthough all transactional I/O happens on the mailbox server, monitoring disk activity is a good way to watch for excessive paging on the CAS. With that in mind, the first counter I recommend monitoring is the LogicalDisk(_Total)/Disk Reads/Sec counter.
If you were monitoring a mailbox server, a high number of disk reads per second would typically be a good thing. In contrast, a client access server doesn't have much reason to access the hard disk — besides the overhead that the Windows operating system caused. Excessive reads usually indicate excessive paging . This often means that the server has insufficient RAM. Microsoft recommends that this counter never exceed 50 on a CAS.
A related counter — LogicalDisk(_Total)\Disk Writes/Sec — is also worth monitoring. Microsoft recommends that this counter also remain at 50 or lower.
Outlook Web Access (OWA) countersThe MSExchange OWA\Average Response Time counter is a good indication of latency that OWA users are experiencing. Microsoft recommends that the average response time always be less than 100 milliseconds. Longer response times usually indicate that the CPU can't keep up with the workload, which is often due to an excessive number of users accessing OWA at a given time.
Remember that excessive CPU usage isn't the only thing that can slow response times. Bandwidth congestion and similar factors can also cause excessive latency.
A similar counter — MSExchange OWA\Average Search Time counter — indicates the average amount of time it takes for searches to complete. Normally, you can't get away with triggering an alert against a counter that is based on averages. However, Microsoft tells us that this counter should never exceed 31000 milliseconds, giving us a threshold value that we can work with when creating an alert.
Offline address book (OAB) countersThe client access server also makes the offline address book available to users. There are two OAB-related counters that you may want to check out.
The MSExchangeFDS:OAB(*)\Download Task Queued counter informs you whether or not the server is waiting for the OAB to be downloaded. A value of one indicates that the server is waiting to download address book data from the mailbox servers but the download has not been executed.
A value of zero indicates that the download has completed. According to Microsoft, this value should always be zero; higher values may indicate communications problems or excessive latency between the CAS and the mailbox server.
The MSExchangeFDS:OAB(*)\Download Tasks Completed counter gives you the number of times that the File Distribution Service (FDS) has downloaded OAB updates since the service started.
It's a bit tricky to set up alerts for this counter, since it should show a steady increase. By default, the FDS downloads updates every eight hours (three times a day). I don't know of a way to trigger an alert if downloads aren't happening, but this counter gets reset if the FDS restarts. Therefore, you might want to create an alert that looks for low values, which would indicate that the service has restarted