Re: [Autoconf] new charter

"Teco Boot" <> Fri, 27 February 2009 17:01 UTC

Return-Path: <>
Received: from localhost (localhost []) by (Postfix) with ESMTP id 6C15528C2B0 for <>; Fri, 27 Feb 2009 09:01:17 -0800 (PST)
X-Virus-Scanned: amavisd-new at
X-Spam-Flag: NO
X-Spam-Score: -0.91
X-Spam-Status: No, score=-0.91 tagged_above=-999 required=5 tests=[AWL=-0.664, BAYES_00=-2.599, HELO_MISMATCH_COM=0.553, J_CHICKENPOX_54=0.6, J_CHICKENPOX_55=0.6, J_CHICKENPOX_56=0.6]
Received: from ([]) by localhost ( []) (amavisd-new, port 10024) with ESMTP id XyCMEosz2Ku3 for <>; Fri, 27 Feb 2009 09:01:16 -0800 (PST)
Received: from (cpsmtpo-eml06.KPNXCHANGE.COM []) by (Postfix) with ESMTP id 54EBE28C268 for <>; Fri, 27 Feb 2009 09:01:16 -0800 (PST)
Received: from ([]) by with Microsoft SMTPSVC(6.0.3790.3959); Fri, 27 Feb 2009 18:01:27 +0100
Received: from M90Teco ([]) by with Microsoft SMTPSVC(6.0.3790.3959); Fri, 27 Feb 2009 18:01:27 +0100
From: "Teco Boot" <>
To: "'Alexandru Petrescu'" <>
References: <> <> <> <000001c99845$1dc56190$595024b0$@nl> <> <1235680887.4585.5.camel@localhost> <> <003901c998cb$42b71e90$c8255bb0$@nl> <>
In-Reply-To: <>
Date: Fri, 27 Feb 2009 18:01:26 +0100
Message-ID: <006801c998fd$06c5bd60$14513820$@nl>
MIME-Version: 1.0
Content-Type: text/plain; charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable
X-Mailer: Microsoft Office Outlook 12.0
Thread-Index: AcmY3q9rlv3+S0D1TIK9du4CoUA55wAD/Kow
Content-Language: nl
X-OriginalArrivalTime: 27 Feb 2009 17:01:27.0128 (UTC) FILETIME=[072C9580:01C998FD]
Subject: Re: [Autoconf] new charter
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: Ad-Hoc Network Autoconfiguration WG discussion list <>
List-Unsubscribe: <>, <>
List-Archive: <>
List-Post: <>
List-Help: <>
List-Subscribe: <>, <>
X-List-Received-Date: Fri, 27 Feb 2009 17:01:17 -0000


|-----Oorspronkelijk bericht-----
|Van: Alexandru Petrescu []
|Verzonden: vrijdag 27 februari 2009 14:24
|Aan: Teco Boot
|CC: 'Alexandru Petrescu';
|Onderwerp: Re: [Autoconf] new charter
|Teco Boot a écrit :
|> Hi Alex,
|> Let's try to be accurate:
|> [skip]
|> |Sorry... in the picture above the addresses are also /128.  It was an
|> |abbreviation for me to show only 2001:db8:1::1/64 assigned to Host1.
|> |The full notation should have been 2001:db8:1::/64 prefix and
|> |2001:db8:1::1/128.  Would the following picture satisfy the need for
|> |/128 addresses?:
|> When prefix::/64 is assigned to a host, it configures a /64 address
|and not
|> an /128 address.
|I'm not sure I understand.
|The prefix::/64 is typically assigned to a link, not to a host.  If a
|host is connected to that link then it configures a /128 address and a
|/64 subnet prefix, both "128" and "64" numbers are visible in its
|I don't understand why the need for /128 prefixes, why isn't the above
|/64-prefix-and-/128address not sufficient?

This is interesting. I meant generating an address in the /64 prefix.
I don't know what is specified in RFCs. I checked behavior on Vista, 
Linux and IOS: 
  o Linux (debian lenny) adds a /128 prefix in the routing table, to 
    the loopback interface, similar to what I propose in my addressing 
    model mail. It also adds a /64 address-prefix to the Ethernet interface
    this is a bit weird, as two interfaces has the same address configured.

  o Vista assigns addresses to the Ethernet interface (in my case),
    and adds /128 prefixes in the routing table. Vista also adds the 
    /64 in the routing table.
  o IOS behavior is as Vista, addresses to interfaces and /128 in 
    routing table.

Details on Linux behavior: the /64 are on Ethernet (eth0) and the /128 
are on loopback (lo).

# ifconfig lo | egrep 'inet6|encap'
lo        Link encap:Local Loopback  
          inet6 addr: ::1/128 Scope:Host
# netstat -6rn | grep 128
::1/128                        ::       Un   0   1    17 lo
2001:db8:1:0:20c:29ff:fee3:bdf5/128 ::  Un   0   1    11 lo
fe80::20c:29ff:fee3:bdf5/128   ::       Un   0   1     3 lo
fe80::20c:29ff:fee3:bdff/128   ::       Un   0   1     0 lo

# ifconfig eth0 | egrep 'inet6|encap'
eth0      Link encap:Ethernet  HWaddr 00:0c:29:e3:bd:f5  
          inet6 addr: 2001:db8:1:0:20c:29ff:fee3:bdf5/64 Scope:Global
          inet6 addr: fe80::20c:29ff:fee3:bdf5/64 Scope:Link
# netstat -6rn | grep 64 
2001:db8:1::/64                ::       UAe  256 0    15 eth0
fe80::/64                      ::       U    256 0     0 eth0
fe80::/64                      ::       U    256 0     0 eth1

Conclusion: I was wrong with my statement. Linux behaves as I mentioned, 
other IPv6 stacks have different characteristics.

|> Routers may generate a /128 prefix-address, and advertize this in the
|> routing domain.
|A host-based route propagated and deleted throughout a domain?  I don't
|see the necessity of doing so.  Assuming the routers are mobile within
|25m ranges then they wouldn't need to change their addresses, thus no
|need to propagate host-based routes.

If the /128 is not propagated, there will be no multi-hop network. In a
MANET, I expect nodes to run a MANET Routing protocol and forward packets.
In ad hoc networks, one (you ?) would say nodes could be hosts or Mobile
Routers acting as hosts. 

|Do you agree we consider routers mobile only within 25m ranges?

Absolutely not. For me, 25km is a reasonable distance! Just 10^3 times the
distance and 10^6 times the power per bit (single hop) or 10^3 times the
power per bit if multi-hop is enabled (and 1000 intermediate nodes....).
Just physical laws here.


|> Some mechanisms should make sure the /128 routing prefix is unique, if
|> required. It is not required if the prefix is meant as anycast
|> routers may use "duplicate prefixes" if this is useful. I think
|anycast is
|> out-of-scope for [Autoconf], but we should be careful when specifying
|> for prefix uniqueness. We should use "SHOULD" instead.
|> Teco.