So here's the setup: exchange 2010 server just using one database with a cas array name with outlook anywhere enabled. Clients are using outlook 2010 from internal and external. When a client goes to setup their profile it works correctly and configures it to use the OA name with the "On fast networks, connect using http first, then connect using TCP/IP" unchecked. It also leaves the "on slow networks, connect using http first, then connect using tcp/ip" checked. The clients are connecting ok, but on the LAN they keep defaulting to HTTPS rather then TCP/IP. I even have network monitor/wireshark running and when the client opens it won't even attempt a mapi connection and then just goes over to HTTPS. If I uncheck the slow networks to use HTTPS rather then TCP/IP the client will then connect using mapi and work just fine. I'm not sure why it would think its on a slow network since everything i found says if outlook thinks its on a connection faster then 128kbps it should think its a fast network. Looking for any ideas at this point why it would be defaulting to this method.

When you in a fast connection, will this issue occur or not? Please run this command to check the OutlookProviderFlags setting, the default is null: Get-OutlookProvider| fl OutlookProviderFlags Related document for you: Set-OutlookProvider http://technet.microsoft.com/en-us/library/bb123683.aspx Thanks, EvanEvan Liu TechNet Community Support

Get-OutlookProvider | fl OutlookProviderFlags OutlookProviderFlags : None OutlookProviderFlags : None OutlookProviderFlags : None So it looks like they are set to what it should be. The system is setup so clients will access content from both intranet and internet. Based on what I have read outlook should think any network connection faster then 128kbps should be considered a fast network. The outlook clients that are having this problem are setup on the local LAN here and have 100Mbps connections.

If so, the settings are right. Are all the users in your environment have this issue? If not all the users have this issue, I suggest you configure your account on another computer to have a try. Thanks, EvanEvan Liu TechNet Community Support

So I ended up trying with a few different machines and a few test accounts and they all gave the same results. What I end up getting now is two connections to mail and two connections to directory according to rpcdiag. But it will end up looking like the below settings: Server Type interface Conn status cas name mail lan HTTP established cas name mail lan TCP/IP established cas name directory lan HTTP established cas name directory lan TCP/IP established

Did you add additional mailbox in this profile? Thanks, EvanEvan Liu TechNet Community Support

Nope no additional mailboxes. Its just one mailbox on the profile.

Are you 100% sure that your client traffic isn't routing outside the org and back in to the CAS Array from external? In other words, do they have the ability to resolve the CAS Array name to a local IP address where an interface on the CASArray member has that IP address bound? I would check what the output of this command is when run from one of your internal Outlook workstations is: Tracert <CASArrayName>

Ok so after some changes we were able to get it working as expected. What happened is internally they are connecting to the cas array name which is going through a hardware load balancer. After checking the config on that device and making some modifications to the persistance profile and connection profile settings the client is connecting on TCP/IP.

Hi, Sounds like the same situation we have occuring in our org - except the problem has been increased when we have moved from XP to Win7. Do you have any further details on exactly what settings were changed for persistance and connection settings on the Load Balancer- e.g what were these set to etc? Or was it more trial and error? Also out of curiosity, what Load Balancer device are you using?

Are you using a reverse proxy if so and you've set it to use basic which is the most common deployment that is expected behavior. Setting the on slow networks does mitigate it but not always I tend to see it with clients as they dock\undock on laptops and routes back out back to the UAG when that happens it terminates your connections and have to re-authenticate due to basic. If you re-design the authentication scheme on the TMG\UAG to NTLM it will stop the behaviro. James Chong MCITP | EA | EMA; MCSE | M+, S+ Security+, Project+, ITIL msexchangetips.blogspot.com

Are you using a reverse proxy if so and you've set it to use basic which is the most common deployment that is expected behavior. Setting the on slow networks does mitigate it but not always I tend to see it with clients as they dock\undock on laptops and routes back out back to the UAG when that happens it terminates your connections and have to re-authenticate due to basic. If you re-design the authentication scheme on the TMG\UAG to NTLM it will stop the behaviro. James Chong MCITP | EA | EMA; MCSE | M+, S+ Security+, Project+, ITIL msexchangetips.blogspot.com HI James, I can find the same issue , LAN users are connect using HTTP? NET-->TMG-->EXG-->Client So i can see 10GB download from the internet using outllok.mydomain.com.au ? How do i allow LAN users to go through the TCP/IP? AS