North American Network Operators Group

Date Prev | Date Next | Date Index | Thread Index | Author Index | Historical

RE: end2end? (was: RE: Where NAT disenfranchises the end-user ... )

  • From: Roeland Meyer
  • Date: Fri Sep 07 16:27:33 2001

|> From: Jon Mansey [mailto:[email protected]]
|> Sent: Friday, September 07, 2001 12:44 PM
|> At 12:31 PM -0700 9/7/01, Roeland Meyer wrote:
|> >|> From: Jon Mansey [mailto:[email protected]]
|> >|> Sent: Friday, September 07, 2001 11:57 AM
|> >|>
|> >|> I seem to be able to connect to port-forwarded services behind my
|> >|> office NAT firewall just fine from my laptop behind my 
|> >|> home NAT box.
|> >|> Whats the problem?
|> >
|> >Can we talk ... using NetMeeting?
|> NM, along with IPsec are examples of apps that dont play well here, 
|> but thats the point, they are apps that have not been written with 
|> the real world in mind, ie that a good proportion of the edge these 
|> days is behind NAT.

NAT is the first and only method (I won't dignify it by calling it a
protocol) that munges the data. Most real protocols only mess with the
envelope and leave the data strictly alone. With NAT, its snafu. Now, if we
had bi-directional transparency, via a NAT proxyd, it wouldn't hurt so bad.
But, such a daemon is impossible to write and, if written, is impossible to

|> Who gives in first here, the app developers (or their marketing 
|> depts) who decide that supporting NAT is important, or the NAT 
|> developers who decide they can fix cuseeme or PPTP by re-writing the 
|> packet data?
|> I am also playing devil's advocate here somewhat, we all know the 
|> real solution to lack of IPv4 space, true end2end, and security lies 
|> with IPv6, right?

1) Feigned IPv4 addr shortages were ameliorated by recovery of legacy IPv4
allocations (/8s). IMHO, too late to prevent us from doing NAT.
2) Routeing table sizes are a routing architecture problem that won't go
away wrt IPv6. They will only get worse there.

<personal opinion>
Whomever shot down IPv6 imbedded routing, needs to be taken out and shot in
turn. The counter-arguments were not convincing, IMHO. I thought it was a
great idea. There is nothing inherently wrong with imbedding the routing
into the protocol itself and it sure helps (a bunch) to standardize things.
</personal opinion>