none
Exchange RPC Counters RRS feed

  • Question

  • On Exchange 2007/ Windows 2008 server, we monitor the following:

    RPC Requests
    Av RPC Latency
    RPC Operations/sec

    Some great links here:

    http://blogs.technet.com/b/exchange/archive/2005/09/28/411674.aspx

    http://technet.microsoft.com/en-us/library/cc540464(EXCHG.80).aspx

    I had some questions:

    1. High RPC Requests - if we start getting high RPC requests values at a certain time, does this mean that there is higher client activity on the server, or that there are more pending requests that haven't been serviced? I'm not sure if this value represents pending activity? How could we tell if this value is because of higher client activity or there are loads of pending requests building up.

    2. High RPC Requests and High RPC Operations/sec - I know that High RPC Requests and static or lowering RPC Ops/sec mean the server can't handle the load, so does High RPC Requests and High RPC Operations/sec mean that the server *can* handle the load?

    3. Is MSExchangeIS\RPC Num. of Slow Packets a good value to indicate delay?

    4. Is there any way to get the above counters per database rather than just for the server as a whole?

    Tuesday, April 12, 2011 6:46 PM

All replies

  • On Tue, 12 Apr 2011 18:46:27 +0000, Smith1974 wrote:
     
    >
    >
    >On Exchange 2007/ Windows 2008 server, we monitor the following:
    >
    >RPC Requests Av RPC Latency RPC Operations/sec
    >
    >Some great links here:
    >
    >http://blogs.technet.com/b/exchange/archive/2005/09/28/411674.aspx
    >
    >http://technet.microsoft.com/en-us/library/cc540464(EXCHG.80).aspx
    >
    >I had some questions:
    >
    >1. High RPC Requests - if we start getting high RPC requests values at a certain time, does this mean that there is higher client activity on the server, or that there are more pending requests that haven't been serviced? I'm not sure if this value represents pending activity? How could we tell if this value is because of higher client activity or there are loads of pending requests building up.
     
    It *may* mean your server is having a problem keeping up with peak
    loads. But what you didn't say was what the values were and for how
    long the numbers are above "normal".
     
    The number here is just an indicvator. The counters you want to watch
    are typically (but not always) those associated with the storage
    subsystem.
     
    >2. High RPC Requests and High RPC Operations/sec - I know that High RPC Requests and static or lowering RPC Ops/sec mean the server can't handle the load, so does High RPC Requests and High RPC Operations/sec mean that the server *can* handle the load?
     
    If the RPC Requests continue climbing until they reach, say, 100, and
    never drop it could also mean you have a "stuck counter". But if both
    counters rise and fall it *may* mean the server has trouble keeping
    up. But those counters, by themselves, are just indicators. Why the
    server can't manage the workload (if that's the problem) is something
    you'll have to use other PerfMon counters to discover.
     
    >3. Is MSExchangeIS\RPC Num. of Slow Packets a good value to indicate delay?
    >
    >4. Is there any way to get the above counters per database rather than just for the server as a whole?
     
    ---
    Rich Matheisen
    MCSE+I, Exchange MVP
     

    --- Rich Matheisen MCSE+I, Exchange MVP
    Tuesday, April 12, 2011 9:50 PM