none
Mail takes exactly 1 minute to be delivered

    Question

  • Hi guys need some help, i have a little problem, my emails take anywhere from 45 secs to 1 minute to be delivered internally, it use to be instant but i have no idea where to begin in troubleshooting this. the only thing that changed was going from 1gb to 10gb nics, which i thought was suppose to be faster o_O

    any ideas

    Saturday, September 21, 2013 9:07 PM

Answers

  • Yes, the very large database, and more important the fact that it's on a single disk volume, is likely the reason for poor local mail delivery.

    If you can spread the mailboxes around to different databases that are on different disk volumes you should see better performance.

    You can test the theory that the single disk is under too much load by looking at disk performance counters Current Disk Queue.

    • Marked as answer by radray Monday, September 23, 2013 4:31 PM
    Sunday, September 22, 2013 10:23 PM

All replies

  • Please provide some more information about your organization's topology.  How many Exchange servers are involved?  How many sites if more than one?  We'll need to know if all your users are on one server or more than one, and what the specific internal mail flow scenario is that you're timing.

    • Marked as answer by radray Sunday, September 22, 2013 6:11 PM
    • Unmarked as answer by radray Sunday, September 22, 2013 6:48 PM
    Sunday, September 22, 2013 5:09 AM
  • No problem,

    2 CAS servers, one configured for internal connectivity and one for external.

    2 Mailbox in a DAG

    1 site, everything is in the same datacentre

    5 databases, however they have 1 database with like 80 percent of the users in it, this database is about 600gb in size

    this was migrated from exchange 2007. same configuration of clustered mailbox and hub cas.

    If you send an email to external it goes instantly, however internal message flow is delayed, seems it went up to 3 minutes today, now prior to this, on exchange 2007, they had the same problem but it was delayed up to 15 minutes, Im guessing its because of the ridiculously large database, but i dont know how to prove this theory

    Thanks again. 

    Sunday, September 22, 2013 6:28 PM
  • Yes, the very large database, and more important the fact that it's on a single disk volume, is likely the reason for poor local mail delivery.

    If you can spread the mailboxes around to different databases that are on different disk volumes you should see better performance.

    You can test the theory that the single disk is under too much load by looking at disk performance counters Current Disk Queue.

    • Marked as answer by radray Monday, September 23, 2013 4:31 PM
    Sunday, September 22, 2013 10:23 PM
  • Thanks a whole lot =D, my disk queue was spiking to 19 at times, i switched over 4 of the 5 databases to the second node and had a noticeable improvement. So i have since had them create multiple stores for different departments and share the load between the two nodes. 

    Thanks again, it was driving me crazy.

    Monday, September 23, 2013 4:36 PM