I'm setting up a quite unusual proxy for a very specific niche case.
I'm allocating (ip,port)
addresses to users, from which they can send and receive UDP and TCP packets.
e.g. say someone is allocated 1.2.3.4:50000
, they can then send packets from that address and receive packets to that address.
Another person might be allocated 1.2.3.4:80
, from which they can then send and receive packets.
Assuming our implementation is sound, and we allocate on IPs we control and ports > 0, are there any security liabilities here?
I know NAT is very similar indeed. I'm wondering as the user will be able to see and choose which of available IPs and port combinations to use, and thus there could be different potentially malicious users with access to port 25, 80, 443, etc, whether this breaks the assumptions of any important protocol.