I have been experiencing a very strange issue with our VPN setup on Windows Server 2008. That server is running as a Xen Virtual Machine. We use it for two purposes, permit our mobile workers to connect to another server hosted somewhere else that only allow that IP, and use it to RDP or ssh to many other virtual machines on the same server.
The server has no performance issue and still a load of memory free. All other virtual machine have no problem whatsoever. Many of those virtual machines have a public IP (web servers) and all their firewall are set to allow only ssh connection or RDP connection from their local interface.
When I am connecting directly with either ssh or RDP to one of the other virtual machines everything works without issue. However, when I am doing so through the VPN after some time the connection just hang, it usually continue after some time (5 or 10 minutes). It seems as more there is network usage more often it happen to a point where it is completely unusable. The worst thing I can do to hang it faster is to actually ping the vpn client IP from the local network, after some time the latency increase until it hang. This happen even if I do RDP to the local ip of the VPN server trough the VPN.
The server report no problem and if I disconnect to the vpn and reconnect right away everything is alright. There is nothing wrong in the VPN server log. I had thought at the beginning that it could have been an issue with the Host server so I try to RDP/ssh directly to the guests and I have experienced no issue while doing this, so it really seems to be a problem with the VPN server on Windows server 2008. Another very weird thing is it does not seems to be an issue if you only do Internet (NAT) without trying to connect to any local IPs.