Re: [Int-area] Continuing the addressing discussion: what is an address anyway?

Toerless Eckert <> Tue, 01 March 2022 19:22 UTC

Return-Path: <>
Received: from localhost (localhost []) by (Postfix) with ESMTP id 93E493A0AE9 for <>; Tue, 1 Mar 2022 11:22:21 -0800 (PST)
X-Virus-Scanned: amavisd-new at
X-Spam-Flag: NO
X-Spam-Score: -0.881
X-Spam-Status: No, score=-0.881 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, HEADER_FROM_DIFFERENT_DOMAINS=0.248, SPF_HELO_NONE=0.001, SPF_NEUTRAL=0.779, T_SCC_BODY_TEXT_LINE=-0.01, URIBL_BLOCKED=0.001] autolearn=no autolearn_force=no
Received: from ([]) by localhost ( []) (amavisd-new, port 10024) with ESMTP id u3PC6elrKtAb for <>; Tue, 1 Mar 2022 11:22:17 -0800 (PST)
Received: from ( [IPv6:2001:638:a000:4134::ffff:40]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by (Postfix) with ESMTPS id 8F5E63A0AF4 for <>; Tue, 1 Mar 2022 11:22:16 -0800 (PST)
Received: from ( []) (using TLSv1.3 with cipher TLS_AES_256_GCM_SHA384 (256/256 bits) key-exchange X25519 server-signature RSA-PSS (4096 bits)) (No client certificate requested) by (Postfix) with ESMTPS id ADCCF58C4AF; Tue, 1 Mar 2022 20:22:09 +0100 (CET)
Received: by (Postfix, from userid 10463) id A3C5D4EA7CB; Tue, 1 Mar 2022 20:22:09 +0100 (CET)
Date: Tue, 01 Mar 2022 20:22:09 +0100
From: Toerless Eckert <>
To: Alexandre Petrescu <>
Message-ID: <>
References: <> <> <> <> <>
MIME-Version: 1.0
Content-Type: text/plain; charset="utf-8"
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
In-Reply-To: <>
Archived-At: <>
Subject: Re: [Int-area] Continuing the addressing discussion: what is an address anyway?
X-Mailman-Version: 2.1.29
Precedence: list
List-Id: IETF Internet Area WG Mailing List <>
List-Unsubscribe: <>, <>
List-Archive: <>
List-Post: <>
List-Help: <>
List-Subscribe: <>, <>
X-List-Received-Date: Tue, 01 Mar 2022 19:22:22 -0000

NAT IMHO has been vilified for a lot of bad instances of NAT from the past.
Now unfortunately, the term is identified with evil only, so maybe we should
find a better term for what IMHO are really useful/beneficial instances.
Maybe "address rewrite".

Non-evil address rewrite IMHO is per-flow-stateless as for example used in IP
with the various SIIT variants.

There are various interesting benficial scenarios for stateless address rewrite, such as 
(but not limited to):

a) the ability to eliminate the need for a single global address plan which eliminates
a lot of administrative work for many type of private networks
b) the ability to build light-weight embedded equipment with hard-burned "generic" addresses -
c) Host stack address stability
d) Traffic steering

a), b), d) are example use-cases for which i've also looked into my FA-IINAS draft
and its address rewrite functions.


On Fri, Jan 28, 2022 at 12:48:59PM +0100, Alexandre Petrescu wrote:
> Sorry, I  take advantage of this valuable public conversation between
> you to mention a point that might be related.
> Le 25/01/2022 à 20:30, Geoff Huston a écrit :
> > [...] various judgemental observations (Like "NAT is evil”, “NBATs break
> > stuff”, etc,) feel free, but they are your constructions, not mine. The
> > issue for me is not judgments of “good” or “bad”, but simply to explore,
> > without overtones of judgement, exactly what an IP
> > address represents in today’s Internet. [...]
> Without jugding, and without thinking others might judge, i.e. to
> qualify as 'good' or 'bad'.
> I do think there might be value in questioning whether there might be
> something inherent in the IP addressing system which might lead to less
> positive consequences.  It is a question on the cause-to-effect dynamics.
> What in the IP addressing system makes it possible that NAT has been
> designed and used largely?  Lack of space in v4 - ok - but is there
> anything more to that problem, now that IPv6 solves the space size
> problem?  Is the fact that NAT kind of probably protection is helping?
> For example, if the IP addressing system had variable length addresses
> (instead of fixed length) - would that make the translation process of
> NAT be unacceptably long, and hence no NAT would be feasible?
> Other than that, what other characteristic of the IP addressing system
> might have an impact on the existence of NATs?
> What other characteristic of the IP addressing system has no impact at
> all on the existence of NATs?  I.e. one could change that characteristic
> but NATs would still be designed.
> Other than NAT and IP addresses, there are other aspects of the current
> Internet addressing that are less desirable.
> For example: the open Internet and its open addressing system leads to a
> need of privacy respecting for the individual; which is good.  At the
> same time, the new privacy rules are not making everyone happy.  Some
> times it goes to large extents.  For example, some addresses of web
> sites are not visible to others _because_ of that privacy ruling.  Not
> all websites in all countries accept to abide to the privacy rules of
> other countries.  Such websites refuse to abide and block access altogether.
> That situation is clearly against the openness of access in the Internet.
> It is not a matter of paying money or not to access data.  Even if one
> pays one is still not given access because one is situated in a country
> of a particular privacy ruling.
> It is a strange situation in which the ruling of privacy is not
> accepted.  Those sites who do not accept to deliver data according to
> the privacy rules do so not because they dont agree with a general
> principle of privacy, but because they dont agree with that particular
> ruling (GDPR in this case) of privacy.
> What is at fault for that situation?
> Is there something in the Internet addressing system at higher layer
> (above IP) that might be qualified as being a little bit in error for
> that lack of access?
> For example, if the 'cookies' used by HTTP involved host names (host
> names are also a sort of addresses) whose structure was agreed locally,
> then there would be more positive view of the generally negative view of
> 'tracking'.  For example, a locally agreed way to identify people is
> generally accepted (license plates, faces, more) but a universal way of
> identification (hostname containing 'Windows' characteristics) might be
> less accepted.
> Alex
> _______________________________________________
> Int-area mailing list