The port isn’t their focus, they’re looking at the protocol that is being used, regardless of the port. The protocol is still visible when not doing deep packet inspection. That’s why there suggesting a socks proxy for Russian citizens, because that uses HTTPS to tunnel traffic, so it wouldn’t be caught up in protocol analysis.
Look up NBAR for the basic idea. Each vendor has their own ‘secret sauce’ implementation, Palo Alto only needs 9 bytes of payload for disambiguation, iirc.
thank you! so it is basically looking at identifiable patterns in the packet flow and matching them to protocols.
i also found this paper about traffic identification interesting.
The port isn’t their focus, they’re looking at the protocol that is being used, regardless of the port. The protocol is still visible when not doing deep packet inspection. That’s why there suggesting a socks proxy for Russian citizens, because that uses HTTPS to tunnel traffic, so it wouldn’t be caught up in protocol analysis.
can you maybe link some ressources on how the protocol used can be detected? i did not know about this and would like to read into it some more :)
Look up NBAR for the basic idea. Each vendor has their own ‘secret sauce’ implementation, Palo Alto only needs 9 bytes of payload for disambiguation, iirc.
thank you! so it is basically looking at identifiable patterns in the packet flow and matching them to protocols. i also found this paper about traffic identification interesting.
Time to up the spoofing game. Maybe some AI-generated traffic to throw off the packet analytics.