Hi @Abela,
It does indeed, TCP is a stateful connection between two endpoints. To ensure that if a connection has failed or otherwise not been terminated properly, the endpoints need to know when to consider a connection dropped. TCP keepalive probe is a way of handling this, if the proxy is busy gathering data and hasn’t sent a packet to the repository’s data mover for a while, it can send a keep alive (this is nothing magic, just part of the TCP specification) to keep the connection alive.
When you have any device performing NAT or a firewall in the middle, this gets more interesting. These gateway devices have a finite amount of network ports available but can carry huge amounts of traffic, so it is more important for these devices to ensure stale connections are closed as soon as possible, if they don’t, they may not have capacity for new traffic. This can cause issues because you might have a TCP keep-alive timer within your application of 300 seconds for example, but the firewall has a keep-alive timer of only 30 seconds. So if the firewall doesn’t see any traffic for 30 seconds, it will consider the connection dead and close the connection.
Realistically most firewalls won’t do this for TCP traffic as it is stateful, whereas for UDP it’s common to see a 30-300 second timer due to them having no stateful control. The firewalls I tend to work with have keep-alives of 3-8 hours. As this issue has just started, it does make sense to look at any potential changes.
A second reason this may occur is if your network topology has changed and you end up with something called asymmetric routing, whereby the traffic is sent via one gateway, but received by another, as the gateways don’t see traffic flowing in both directions they can determine the traffic is invalid and close the session. Performing traceroutes on both endpoints to the other is a great way to detect this.
I’m also making the assumption here that you’re not using NAT between these endpoints (if one side is 10.0.0.1 and the other side is 10.0.1.1, they see these IP addresses of each other and not being changed to a router/firewall’s IP address)
We’re going down a very specific rabbit hole here and it may be the complete wrong thread to pull at, so I’d suggest following the advice of the others as well and supply some logs, do you have Veeam Support? If so they’ll be able to spend some time with you going through such common issues.
Good Luck! 
