I have a much easier time understanding PAT than NAT addressing. My question is on how NAT identifies an incoming packet target adress on the internal network.
PAT uses ports as identifiers, but what about NAT?
Add a comment
|
2 Answers
With NAT there are 3 possibilities:
1.) multiple private addresses in local network (inside local) get mapped to single public IP (inside global) --> This is PAT or overloaded NAT private IPs are distinguished by port from outside
2.) single private to single public -> Basic NAT, it has a table of translations, this is called Static NAT
3.) multiple private to pool of public IPs -> this is called Dynamic NAT, it will choose some public IP from pool for every host
If you want to do true address-only NAT without getting port translation involved, you must have the same number of addresses on one side as the other. So the first public address matches to the first private address, the second to the second, the third to the third, and so on, to the end of the address pool.
-
Not entirely sure i follow. You mean that in order for true NAT it has to be a 1:1 relationship between public & internal IP?– ErikAug 10, 2019 at 7:28
-
1@Erik Yes. If you have fewer public IP addresses than internal IP addresses, you must do what the standards call NAPT (which in Cisco-land is called "PAT").– SpiffAug 10, 2019 at 7:33
-
1So basically today, NAT isn't used much in home environments becuase the IPv4 shortage but rather PAT is used. Just becuase there arent enough addresses to go around for a real NAT environment. Awesome! Thank you for helping me getting a grip on this :D– ErikAug 10, 2019 at 7:54
-
@Erik: Yes, but it's still called NAT by many, even though it involves port translation. Different places and even different IETF RFCs use different terminology -- what you call "NAT" vs "PAT"/"NAPT" is also very often named "1:1 NAT" vs "cone NAT". That's probably a source of confusion. Aug 10, 2019 at 8:14