Exchange 2010 monitoring

There are scores of counters on Exchange to be aware of, but these give a good high level representation of the infrastructure:

• %Processor Time: < 75% (average)
• Processor Queue Length < 5
• Memory Available > 100 MB
• Memory Transition Pages/Sec < 100, spikes < 1000
• Memory Page Reads/Sec < 100
• .net clr mem% time in Global Catalog < 10%
• TCPv4 /v6 connection resets should be low
• LDAP time < 50 ms, max < 100ms
  

Additional Counters for CAS

• OWA average search time < 5000ms
• ASP.net App restarts / Worker process restarts / request wait = 0
• RPC Client Access / Latency < 250ms
• RPC Client Backoff /sec = low numbers, should no increase
  

Additional Counters for Hub Transport:

• Average Disk Reads/Sec < 20ms, spike < 50
• Average Disk Writes/Sec < 20 ms, spike < 50
• Aggregate Delivery Queue Length < 3000, spike < 5000
• Remote Delivery < 250
• Submission Queue < 100
• Retry Delivery Queue <100
• Largest Delivery Queue < 200
  

Additional Counters for Mailbox:

• I/O reads avg latency < 20ms < 100ms
• Page faults stalls/sec 0 (disk I/O)
• RPC requests < 60
• RPC avg latency < 10ms
• Database Instances / Log gen check < 500, 20-30 per DB
  

Additional Counters for UM:

• Message success processed /hour > 95%
• Call answer queued messages < 50
• Direct access failures =0
• Operations > 6 seconds =0