I just completed an Exchange 2007 to Exchange 2013 CU5 migration for a client that has about 80 mailboxes (though currently the Exch2007 server is still in place with no mailboxes on it). Everything went relatively well. About the only show stopping issue I ran into was mobile device non-access for users that didn't have Inheritable Perms set in Advanced settings on their AD object.
However, I'm also getting complaints of slow Outlook performance on three physical terminal servers with Outlook 2010 running in non-cached mode. Specifically users are complaining that when they reply to email, it can take 3-5 seconds for the reply window to pop-up. I verified this by logging in as one of the users onto one of the terminal servers. Of course if I turn on cached mode, replies don't lag on the TS, but that isn't optimal in this environment with this amount of email. Outlook installed and run directly on the Exchange server doesn't experience the same lag in non-cached mode. However, if I turn on the "Show Network Connectivity Changes" notification (right-click Outlook tray icon, check it) Outlook continually pops up messages saying it has lost connectivity/been restored. Because I just learned in the last hour that connectivity to Exchange 2013 is ONLY RPC/HTTPS now, I don't know if this continual HTTPS drop/reconnect is normal. Again, I'm seeing this in Outlook running directly on the Exchange server.
Exchange 2013 is running on a Win2012 Standard virtual server on a host running VMWare 5.5. The Exchange server has been allocated 16GB of memory, 8 CPU cores, a vmxnet3 adaptor, VMware Tools is installed, High Latency setting has been applied with fully reserved memory and CPU. IPv6 is disabled on the server via the registry. The server is fully patched. Throughput tests from the terminal servers to the Exchange server shows results typical of GB speeds. OWA works fine and is peppy. I also created a new throttling policy in Exchange with most of the settings set to unlimited and applied it to the mailbox I'm testing with.
Does anyone have any insight on what to look at to troubleshoot this, or is this the new standard for RPC/HTTPS only connections in non-cached mode?