It’s a little more complicated than that. When a user initiates a Lync call with one other user, and they are both inside the network, then the traffic is routed between the 2 users. It’s similar to the way a SIP or SCCP (VoIP) phone call is made. The server builds the connection between the 2 users and then is left out of it. When there are 3 or more users, OR there is media sharing (not simple desktop sharing), then all traffic routes through the Front End servers in Valley Forge. Internal users connect through the global MPLS network, and external users connect via the Internet through a reverse proxy and edge server in Valley Forge. Any MPLS traffic (not sites connected via VPN) should be QoS’ed based on the voice rules, but any other traffic would be “best-effort”
I hope this clears thing up a little on how to potentially troubleshoot quality on Skype calls. Please try to make sure that the clients are updating Link (Skype for Business) to the latest version as Microsoft is always making improvements.
As for SRC, we can always ask them to see if there’s any network issues. They typically see things in their monitors before we do, but sometimes, errors that aren’t high enough to set off and alarm in the monitors can cause unusual issues with services like VoIP and Skype for Business.