eG Monitoring
 

Measures reported by XchgClientTest

Monitoring the RPC activity to an Exchange server and the responsiveness of the server to RPC requests can provide an indication of user satisfaction levels with the performance of the Exchange server. Foreground RPCs happen during interactions of Outlook clients with the Exchange server, and any slow down or failure of these RPCs will be directly visible to users of the Exchange server. Background RPCs are caused by "behind-the-scene" activities internal to the Exchange server.

The XchgClientTest monitors the performance of RPC activities on the Exchange server. Since RPC related metrics are available only from Exchange Server 2003 onwards, this test is only relevant for Exchange 2003 or higher versions.

Measurement Description Measurement Unit Interpretation
Rpc_attempts The rate at which RPC calls were attempted to the Exchange server during the last measurement period. Atttempts/sec This metric provides an indicator of the RPC workload on the server.
Rpc_failures This metric is the rate of failed RPCs to the Exchange server during the last measurement period. Failures/Sec Typically, this value should be low
Rpc_successes The rate of successful RPC requests handled by the Exchange server during the last measurement period. Successes/Sec  
Foreground_rpc_failures This metric is the client-reported rate of failed foreground RPCs during the last measurement period. Failures/sec Typically, this value should be low.
Foreground_rpc_successes Shows the client-reported rate of successful foreground RPCs handled by the Exchange server during the last measurement period. Successes/Sec  
Rpc_success_ratio This metric is one measure of client satisfaction levels with the Exchange server. This metric is the ratio of the foreground RPC successes to the total number of foreground RPCs attempted during the last measurement period, expressed as a percentage. % The closer this value is to 100, the better the client satisfaction level.
Rpc_latency_2secs Shows the client-reported rate of successful RPCs during the last measurement period with latencies > 2 seconds. Rpcs/sec  
Rpc_latency_5secs Shows the client-reported rate of successful RPCs during the last measurement period with latencies > 5 seconds. Rpcs/sec  
Rpc_latency_10secs Shows the client-reported rate of successful RPCs during the last measurement period with latencies > 10 seconds. Rpcs/sec  
Fast_rpc_ratio This metric indicates whether client RPCs are happening fast or not. % This metric is another key measure of client performance. This metric is computed as the ratio of successful client RPCs with latency less than 2 seconds to the total number of successful RPCs, expressed as a percentage. Hence, the value of this metric indicates the percentage of client RPCs that are taking more than 2 seconds. A value closer to 100 indicates better RPC performance.
Server_to_busy_per_sec Indicates the rate at which the RPCs failed due to the server too busy ROC error. Error/sec Ideally, the value of this measure should be 0. A higher value may indicate the exhaustion of RPC threads or occurence of client throttling for clients running versions of Outlook earlier than Office Outlook 2007.
Server_to_busy Indicates the number of failed RPCs due to the server too busy ROC error. Number Ideally, the value of this measure should be 0.
Slow_qp_thread Indicates the number of query processor threads currently running queries that are not optimized. Number Ideally, the value of this measure should be low. The query processor accepts, selects and then executes the query. The Exchange information store includes a query processor that performs property search of datas in the information store. It retrives result from the Property Store DB and the query Components. The query processor will load balance request if more than one query component (mirrored) exists within the same Index Partition.
Slow_search_thread Indicates the number of search threads currently running queries that are not optimized. Number Ideally, the value of this measure should be low.
Rpc_client_backoff Indicates the rate at which client backoffs are occurring. Backoffs/sec Ideally, the value of this measure should be 0. A higher value of this measure is a clear indicator of the server incurring a higher load resulting in an increase in overall averaged RPC latencies thus causing client throttling to occur.

This can also occur when certain client user actions are being performed. Depending on what the client is doing and the rate at which RPC operations are occurring, it may be normal to see backoffs occurring.that there is an increase in overall RPC latencies or occurence of client throttling.