Wow, this is very good feedback. I don't think this is documented anywhere, but I'll put this onto the forums for everyone else. Yes, the keep alive is handled by tcp port 443 which is a control channel while data is sent over UDP 4500. If there is a NAT involved, then it is likely the keep alive is responded by the NAT and never reaching the PCS device. If this is correct, then the PCS device has its own keep alive. The logic is the following: After 300 seconds (5 minutes), PCS will send a keep alive to the client. If not response is received it will send a keep alive every 75 seconds up to 9 times for a total of 675 seconds. If no response is received after 300 + 675 = 975 seconds, then the PCS device will drop the connection.
... View more