S2S IPSEC VPN throughput/performance issue with Exchange

We are seeing some performance issues on our IPSEC VPN connected WAN.  

1. We have a remote office connected through a S2S IPSEC VPN.  
2. Both WAN connections are Comcast Business 150/25 Mbps
3. We are both on SG210 appliances - latest and greatest firmware as of last week.
4. 70% Remote branch users are on branch  local 2008R2 Terminal Servers running Outlook and access Exchange Server at Corporate
   1.  Outlook is NOT using cached mode because of Terminal Services
5. During the day, performance of Outlook will degrade to a complete standstill for TS users
   
   1. No spikes on either firewall in traffic - traffic patterns look consistent day-to-day and week-to-week
6. Rebooting corporate FW  clears up problem until it degrades again

Other factors/information

1. We upgraded  to Exchange 2013 and put a server/DAG at the remote site to give them local performance improvement
2. When we switched to the new DAG, performance became worse for the users and Exchange threw a bunch of errors related to communication timing out between the servers
3. We assume something is filtering the traffic but not sure what - everything across the VPN should not be filtered?
4. Remote access to servers and resources seems to be fine
5. There is a setting in Exchange 2013 for Outlook Anywhere that when you click on this setting, you should get a pretty instantaneous reponse
   
   1. When we click on this setting for the DAG at the remote office, the information either times out or it takes 3-5 minutes on average to return the configuration information
6. Ping times between the servers are solid at 25-40ms

I have made sure all DNS records are proper and all forwarding etc is pointing proper

To troubleshoot/rule out filtering I have turned off ALL filtering I can find - web filtering, IPS, Application, QoS

I have created rules in the firewalls to allow  for  RemoteLAN <-> CorpLAN ANY protocols

My technical consultant (very bright)  says he has Comcast clients with LESS bandwidth doing the exact same thing with MORE users connecting and MORE data being pushed between sites.   Difference being they are not on SOPHOS

Right now the DAG is sitting as a shadow/copy because we cannot make it the primary DB for my remote users because of problems.  

Help!