I'm investigating reasons why several Outlook clients in non-cached mode are having delays communicating... resulting in Outlook hanging.
My suspicion is that a Netscaler, or intermediate device (? examples wanted) may be interfering with Exchange - Outlook communication.
I am terminating the HTTPs session at the netscaler, then using 80 to the backend (SSL Offload), and think this might benefit.
Is there any documentation that would support the theory that Nagle + Outlook shouldn't mix?
The only counter-argument I can think of is Activesync and long-polling. For this we have a different hostname in Autodiscover, so the features and functionality shouldn't be impacted for these devices, and my policy can be targeted to Outlook fat clients.