locked
Hub mailflow not load balancing RRS feed

  • Question

  • Hello,

    We have an AD site that has 10 Single Copy Clusters exchange 2007, 4 hub servers 2007 and two CAS servers 2007. We have noticed that mail flow is not uniformly distributed among all the hub servers but instead one hub server gets most of the traffic in and out. For instance, if there is 100K mail a day transferred in the AD site, one HUB server handles 70K while the rest 30K is split between 4 HUB servers. When we set these hub servers we did not make any modification and my assumption is that the traffic would be load balanced at the AD site level. The 4 hub servers are all configured the same but I am not sure why traffic defaults to one hub the most. Any ideas would be helpful.

     

    Monday, May 9, 2011 11:54 PM

Answers

  • Hi Mike,

    Any update for your issue?

    Regards!
    Gavin
    Please remember to click “Mark as Answer” on the post that helps you, and to click “Unmark as Answer” if a marked post does not actually answer your question. This can be beneficial to other community members reading the thread.
    • Marked as answer by Mike Logan Tuesday, September 20, 2011 5:51 PM
    Tuesday, May 17, 2011 8:32 AM

All replies

  • Hi,

    There is two situations for load balancing for inbound messages from internal servers and for internet messages in HUB Transport Servers. There is always a debate for truly load balancing in HUB transport servers, So Microsoft recommend customers to use Hardware LB which give much more load balancing if you are going to receive huge email from third party applications, POP source.

    There are lots of things which affect redundancy for HUB servers, Please have a look of below KB in detail.

    Load Balancing and Fault Tolerance for Transport Servers

    http://technet.microsoft.com/en-us/library/bb267003(EXCHG.80).aspx 


    Anil
    Tuesday, May 10, 2011 9:03 AM
  • Is this only from internet or do you have other Exchange servers in other AD sites?

    is mail coming in from Internet targeting your HUB servers directly or do youhave anything between internet and your HUB servers?

    if target directly, then you must play with DNS and MX records. if there is some gateway between,then you must configure the gateway to distribute mail correclty.

     


    lasse at humandata dot se, http://anewmessagehasarrived.blogspot.com
    Tuesday, May 10, 2011 11:04 AM
  • This is for internal mail only. We have three sites but all traffic goes through this site. Mail coming from internet does not hit our hub server directly.
    Tuesday, May 10, 2011 12:51 PM
  • This is for internal mail only. We have three sites but all traffic goes through this site. Mail coming from internet does not hit our hub server directly.

    Stop the Transport service on the one HT that seems to be doing all the work for ahwile and see how the load is distributed across the remaining hubs. In theory, the hubs should be used in a round-robin fashion.
    Tuesday, May 10, 2011 1:16 PM
  • Did that and mail is more or less equally distibuted between the three which is what we need. Once I start the trasport service on the other one it starts processing more traffic which I find odd.
    Tuesday, May 10, 2011 3:15 PM
  • Issue a Set-MailboxServer for the mailbox servers and see if someone set the SubmissionServerOverrideList

     

    If they did, then stopping that HT should cause all submissions to stop, but still worth checking.

     

    Tuesday, May 10, 2011 3:21 PM
  • Did a get-mailboxserver for the entire org and checked for SubmissionServerOverrideList. All undefined "{}" and nothing is hard coded.
    Tuesday, May 10, 2011 3:28 PM
  • Mike, I have seen this kind of situation in many Exchange deployments where Exchange HTs are load balanced using WNLB. Is this the case for your environment too? If yes then I am sure there is nothing much that you can do with this. During recent publication MS has started recommending using HLBs instead of WNLB for load balancing the servers. I am sorry that I do not have a document handy to link right now but there is a technet article which would help you the reasons for NLB not working in case of HT servers. Read it here http://technet.microsoft.com/en-us/library/bb267003(EXCHG.80).aspx Hope that helps.
    Milind Naphade | MCTS:M (Exchange 2007 and 2010) | http://www.msexchangegeek.com RSS Feed
    Tuesday, May 10, 2011 3:51 PM
  • Milind,

    Indeed we have NLB configured on this hubs so that SMTP email from applications that use anonymous authentication so we specified some receive connectors and used WNLB. However, this should not affect mapi traffic.

    Tuesday, May 10, 2011 4:08 PM
  • have you configured transport dumspter equally on the HUB servers?

     


    lasse at humandata dot se, http://anewmessagehasarrived.blogspot.com
    Tuesday, May 10, 2011 8:14 PM
  • Do we need to in an SCC environment? However, we have it configured to 18 MB/SG.
    Wednesday, May 11, 2011 1:06 PM
  • Do we need to in an SCC environment? However, we have it configured to 18 MB/SG.


    18 is the default.

    If you stop the server in the WNLB console, but leave the transport services running, does it still happen?

     

    Wednesday, May 11, 2011 1:30 PM
  • Hi Mike,

    Any update for your issue?

    Regards!
    Gavin
    Please remember to click “Mark as Answer” on the post that helps you, and to click “Unmark as Answer” if a marked post does not actually answer your question. This can be beneficial to other community members reading the thread.
    • Marked as answer by Mike Logan Tuesday, September 20, 2011 5:51 PM
    Tuesday, May 17, 2011 8:32 AM