(ngtrans) Re: again: draft-blanchet-ipngwg-testadd-00.txt

Robert Elz <kre@munnari.OZ.AU> Wed, 27 June 2001 18:55 UTC

From: Robert Elz <kre@munnari.OZ.AU>
To: Alain Durand <Alain.Durand@sun.com>
cc: Marc Blanchet <Marc.Blanchet@viagenie.qc.ca>, ngtrans@sunroof.eng.sun.com
Subject: (ngtrans) Re: again: draft-blanchet-ipngwg-testadd-00.txt
In-Reply-To: <5.1.0.14.0.20010627104811.00b0bfa0@jurassic>
References: <5.1.0.14.0.20010627104811.00b0bfa0@jurassic> <3297.993656515@brandenburg.cs.mu.OZ.AU>
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Date: Thu, 28 Jun 2001 01:53:59 +0700
Message-ID: <4069.993668039@brandenburg.cs.mu.OZ.AU>
Sender: owner-ngtrans@sunroof.eng.sun.com
Precedence: bulk
Reply-To: Robert Elz <kre@munnari.OZ.AU>

    Date:        Wed, 27 Jun 2001 11:00:20 -0700
    From:        Alain Durand <Alain.Durand@sun.com>
    Message-ID:  <5.1.0.14.0.20010627104811.00b0bfa0@jurassic>

  | I'll keep the discussion in NGtrans.

Well, OK, though I think that the documentation of IPv6 (I mean end-user
doc) is just as important as the code, and not at all related to
transition, so I'm not sure why it would be here.   But where the
discussion takes place doesn't matter much to me.

  | The reason why I'd like to have this prefix under 3ffe::/16 is that
  | we already have the delegation here, thus the is no need to ask for
  | something new.

That's bogus.   We don't have to ask - we (that is, a subset of us, which
doesn't include me really) invented the whole thing, and have already set
aside blocks for various purposes.   We can do the same again.   All we have
to ask for is that the assignment be recorded - that's no big burden.

  | In my opinion, the main reason to want something outside of 3ffe::/16 would
  | be if the 6bone prefix was to be reclaimed and reassigned,

No, that's not it (though I guess it is a small side issue for the far
future).

The reason is so that addresses used in doc simply don't look at all like
addresses used in real life - that is, it is trivial to distinguish one
from another.  A prefix carved out of 3ffe::/16 (even if it is 3ffe:ffff::/32)
is pretty similar to any other 6bone prefix.  It looks just the same,
people will copy the address from the doc, and configure it.   That's what
I really want to avoid happening (or at least, when they do, I'd like the
software to be able to say "invalid address" as soon as they attempt it).

  | As about prefix length, /9 seems to me overkill and a real waste of address
  | space.

There's something to that - but I think doc is worth it.  It allows all
levels of the hierarchy to be used in examples, to give config examples of
how NLAs talk to other NLAs and filter addresses, ...

The bigger concern, than the size, is that we might one day want fe00::/9
for more kinds of "local" addresses (to sit next to site local and link local).
That one when I initially thought about this I decided wasn't worth worrying
about - the alternative would be to use fc00::/8 for this, and leave fe00::/9
vacant.  That looked backwards, and to actually be a waste.

kre

(ngtrans) Re: again: draft-blanchet-ipngwg-testad… Robert Elz