04-02-2020 10:49 PM
this is how pbf works
consider how you redirect packets out to the internet: a syn packet goes out to establish a session: does it need to take a route or go through pbf? all we can see at this point is port 443
if pbf is set for ports, it will send via pbf, if it's set for applications, it can't know at this point, so the packet is set out through regular route
so the syn/ack comes back and another ack is sent out via the regular route, still no clue what application we're talking and we're already 3 packets in
then comes payload and app-id can go to work. once a rudimentary application is identified, pbf can kick in. for this session however it's already late but app-id leanrs this session and stores it in the app cache so it can apply pbf on the next connection to the same source/destination/port combination
this is why it's recommended not to use applications in pbf, and why only some app-ids are available to begin with (because these are more easily identifiable and can use this 'bypass' without too many false positives)
hope this makes sense
