Difference between NAT and NAPT?
Marko Vojinovic
vvmarko at gmail.com
Wed Apr 29 21:49:13 UTC 2009
On Tuesday 28 April 2009 07:51, Nifty Fedora Mitch wrote:
> On Sun, Apr 26, 2009 at 05:51:52PM +0200, Marko Vojinovic wrote:
> > In short, what is the difference? Are there any (dis)advantages of
> > using one over the other?
>
> Put your subject line in a search engine like Google.
>
> http://www.dcs.gla.ac.uk/~lewis/networkpages/m05s09NAT.htm
>
> For most "mortals" NAT is just fine.
> NAPT may be needed on a large private network but
> the hardware/ software has to work harder and thus
> may cost more.
Well, I was already beginning to worry if my post made it to the list at
all. :-)
Of course, I did do some research on the subject, but all that I found was
described in a very general way, and consequently vague. So I was hoping to
start a conversation with someone knowledgeable, in order to get more
concrete answers.
My setup consists of three to five computers and a small wireless router, with
an adsl uplink utilizing a dynamic public IP address (just a single one, the
m=1 case in the article you quoted). What I would like to understand better
is the following:
* Why does my ISP's router manual insists on using NAPT over NAT? The ISP tech
support admitted to not understand why and have no explanation, but
nevertheless they suggested that I set up the router as the manual says. Is
there a general well-known reason for insisting on such a setup?
* Is there a performance penalty in using NAPT over NAT? Packets have to be
altered and reassembled in both cases, so should I really expect any notable
time difference here?
* Given my setup from above, is there a serious need to use NAPT over NAT? If
yes, why? If not, why not? (note: I consider muself just a simple mortal with
a small home network, nothing too fancy)
* I understood that NAT is about mangling the source IP address of the packet
so one could push more local IP addresses through less public ones. I fail to
understand the further gain of mangling tcp/udp port numbers? Can you provide
an example situation where NAPT works and NAT doesn't, so I can visualize the
difference in packet travel?
* Is it probable that the NAT setups I have created in the past (typically on
Linux machines playing as routers, using mostly firestarter built-in NAT
support) were actually NAPT setups, while I wasn't explicitly aware of the
difference? IOW, is it maybe usual to say/write NAT in software manuals while
actually meaning NAPT instead?
I tried to do my homework here, but these questions somehow just weren't quite
answered in any NAT vs NAPT articles I could find on the net. I would
appreciate any hints in understanding all this a bit better.
Thanks, :-)
Marko
More information about the fedora-list
mailing list