Exchange 2010 extremely slow DS connections

ssjaronx4ssjaronx4 Member Posts: 37 ■■□□□□□□□□
Hi Guys,

I'm having problems with an Exchange 2010 deployment, 2007 - 2010. I have setup the CAS and MBX server and it works fine except the directory connections are extremely slow. When you use Outlook to connect to any mailbox on the 2010 servers, the directory connections are around 8000 -10000 and I cant figure out why.

The servers are W2K8 R2 and the firewall is disabled and Exchange is SP1. The servers are running on VMware ESX 4. The Exchange 2007 server sit on exactly the same box and they do not show any problems.

I have fixed an issue with RPCSS on the DC's and cant see anything else wrong with them. I have re-run Schema, AD, Domain prep and there dont appear to be any issues there. I've disabled TCP offline and turned of IPV6 which hasnt changed much. I do have autodiscover pointing back to the old HT\CAS server ATM but dont see why this should cause any problems (am I missing something here?).

Just looking for some ideas as this is doing my head in =).

Comments

  • ssjaronx4ssjaronx4 Member Posts: 37 ■■□□□□□□□□
    Hay Guys,

    This morning I figured out what was wrong. The slow DS connections were being caused by some issues with a specific DC.

    There is a parameter within the Set-ExchangeServer cmdlet which allows you to exclude domain controllers from use by Exchange: -StaticExcludedDomainControllers. I used this parameter to force Exchange to use 1 DC at a time. I found that whenever the offending DC was in use the DS connections were around 10000 ms. When any other DC in the environment was targeted, connections were 0-3ms. By excluding the DC from the list of DC’s i have managed to get the Exchange environment working as expected.

    When you use this command you will see the list of DC’s on the properties of the CAS server in the EMC change. Here is an example of how to exclude DC’s:

    Set-ExchangeServer -identity CAS.Example.co.uk -StaticExcludedDomainControllers:dc1.Example.co.uk,dc2.example.co.uk

    To reset the list, use the following command:

    Set-ExchangeServer -identity CAS.example.co.uk -StaticExcludedDomainControllers:$null

    Guess this explains some other wierd performance issues within this environment. I'll look into the problem with the DC tomorrow and may consider rebuilding. Might be a good excuse to start getting the DC's up to 2008 R2.
  • ssjaronx4ssjaronx4 Member Posts: 37 ■■□□□□□□□□
    Quick update.

    Found the culpret. It was server side scaling in 2008 causing the issue. Used the following command to disable it on all 2008 R2 VM's:

    netsh interface tcp set global autotuning=disabled

    Everything in the environment is now speeding along compared to before.
  • MishraMishra Member Posts: 2,468 ■■■■□□□□□□
    Are you seeing improved performance from your functioning DCs?
    My blog http://www.calegp.com

    You may learn something!
  • ssjaronx4ssjaronx4 Member Posts: 37 ■■□□□□□□□□
    Not at all. We have 4 DC's in the environment, 2 virtual and 2 physical. There are 3 sites, 1 physical dc in the main office, 2 virtual dc's in one data center and a physical dc in the dr data center.

    The DC having the issue was the one in the main office.Its a wierd one as none of the other DC's has any problems what so ever but this one DC had issues only when the 2008 VMs were trying to communicate with it. A physical box running 2008 had no issues either.

    The issue only occured when the 2008 VM's tried to perform functions against the DC in the main office e.g pulling a file from the DC was very very slow (200kb a sec on a gigabit line) or doing DS lookups. You could copy a file from the DC to the VMs at full speed.

    We tried different virtual NICs and other stuff but none of it had any effect. A colleague has seen similar behaviour in a different senario with virtualised servers so we thought we would try disabling server side scaling and it worked like a charm.
Sign In or Register to comment.