1
votes

Several months ago a second exchange server was installed in our domain at a data center, wear-as the first server is located at the local office location.
Both Servers are exchange 2010 SP2 and everybody user outlook 2010 and both have a separate UCC SSL certificate for their own domain.
Both servers have CAS, Hub and Mailbox server roles.
Half of the existing staff members were moved to the new server and are sending and receiving emails via the new server (trusted domain).
The biggest issue for the staff members on the new server is that the shared calendars work very slow. often when opening a shared calendar or setting up a meeting outlook just gets stuck and after half a minute or so and pops up a message "Outlook is trying to retrieve data from Microsoft Exchange server server name.domain." - this mainly happens when attempting to add rooms from the address book and then pops up an error "the operation failed".
Sometimes a message that pops up is:

The address list cannot be displayed. 
The connection to Microsoft exchange is unavailable. 
Outlook must be online or connected to complete this action. 

This is an issue that everybody experiences from the new exchange server. when comparing the the configurations of the two servers noticed that the local auto-discover url of the original EXCH SRV was https://mail.domain.com/Autodiscover/Autodiscover.xml and not https://'netbiosname'/Autodiscover/Autodiscover.xml.
After changing the internal URL for the new server to be the same as the external the performance improved a lot, but there are still many of those popup messages and "spinning wheels". And it seems like lately the problem is just getting as bad as it was before the change.
There are no errors or warnings in event viewer.
I have setup fiddler on my pc and i noticed a number of errors:

  1. There are many 401's when connecting to /ews/exchange.asmx. After several 401, there is a 200.
  2. Outlook connects to the EXCHserver1 autodiscover and not EXCHserverNew - and also gets several 401's before getting to the 200.

When working on an account that is located on EXCHserver1 in fiddler you can just see direct tunnels to mail.domain.com.
When I loaded a mailbox at the data center where the EXCHserverNew is located the performance was much faster but still not 100%, and there were still the same errors showing in fiddler.
But when I tried a EXCHserver1 account at the data center, everything worked just fine.
My last resort to solve this is to move the New exchange server to the local office, which I want to avoid since it wont solve the problem 100%.

1

1 Answers

1
votes

I ended up opening a case with Microsoft, and after two weeks of tests with them several things were done to improve the severity of the issue:

  1. Increased the ThrottlingPolicy for the EXCHserverNew cmdlet in EMS
  2. Disabled TCP/IP metrics 'TCP Chimney', 'RSS', 'Autotuning' & ‘NetDMA'
  3. Disabled IPv4 Checksum Offload", "Large Send Offload Version 2 (IPv4)", "Large Send Offload Version 2 (IPv6)", "TCP Checksum Offload (IPv4)", "TCP Checksum Offload (IPv6)", "UDP Checksum Offload (IPv4)", "UDP Checksum Offload (IPv6)", on the NIC card level of the EXCHserverNew
  4. Changed the Hosts file on the clients machines to point ot the local IP on the exchange server.

This improved the issue a lot, but outlook client calender still takes a second or two to load or open the OAB.