Hi All
Background: I run an environment of around 8 RDS 2012R2 Session hosts on 3 physical ESXI hosts. I have around 120 users logged onto RDS using Office on the session hosts as well as a 3rd party broker software. The physical hosts and vms easily have enough resource (CPU/Memory/IOPS etc) to handle the load. I run Office 2013 on majority of session hosts but a couple now have Office 2016 (session hosts are split between collections and security groups). The RDS environment/DCs/Exchange servers and all other vm servers run on the same physical ESXI hosts/same local network connection.
Exchange VM has 20GB Mem/2 CPUs and plenty of disk space. RDS Session Hosts have 20GB Mem and 4 vCPUs. During peak hours RDS severs and Exchange peak at around 60-70% of resources max!
Applications are deployed via remoteapp via RDWeb but i have tested the issue below by rdp direct to a session host.
Exchange version is 2013 CU 11
So now to the problem. Ongoing, we have been experiencing slow RDS Outlook performance. Outlook runs in on-line mode and intermittently navigating between emails/clicking emails, displaying a preview of that email can sometimes have a delay
of 5-10 seconds. 10 minutes later and the performance has gone back to normal. It appears to only affect clicking between emails because if i then compose an email or click into menus, performance is normal.
I found this article and tried the reg key (HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\Tcpip\Parameters\Interfaces) but this hasn't fixed the issue. If I check connection status in Outlook when performance is slow, Avg Resp times are well below 100 ms.
I am really at a loss to what is causing the issue as everything else performs fine. I have tested local Outlook on my PC and performance is good. I am starting to look into changing to cached mode in RDS which I have tested and "resolves the issue" but then I have to think about local large .ost's and roaming profiles and backups and multiple session hosts etc which will be a nightmare.