[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: IPv6CP - rfc 2472 --- why only IID, why not prefix too

Hi Greg,

Thanks for the kind & prompt response.
What you mentioned is correct; but this kind of provision in IPv6 framework
has resulted in more complex implementations.

Eg., if PPP server would have allocated
one prefix to PPP client (in IPv6CP) then the client would become a globally routable
host; and the PPP client might use this prefix as per his requirements;
and multiple prefix allotment could also take place in the same way as
defined in IPv6 RFCs!

But since things are not defined in this way, PPP client and
server both have to have DHCPv6 client/server implemented
on them; making both the systems (PPP client & server)
more complex!

Thanks & regards

From: Greg Daley <greg.daley@xxxxxxxxxxxxxxxxx>
Reply-To: greg.daley@xxxxxxxxxxxxxxxxx
To: Rajan Srivastava <rajan_srivastava@xxxxxxxxxxx>
CC: ipv6@xxxxxxxx
Subject: Re: IPv6CP - rfc 2472 --- why only IID, why not prefix too
Date: Thu, 25 Nov 2004 13:09:05 +1100

Hi Rajan,

Rajan Srivastava wrote:

In rfc 2472, it is provisioned that PPP server will negotiate
interface identifier with PPP client, server will not pass on network prefix;
and prefix delegation is left for DHCPv6 etc.

What is rationale behind this strategy?

IP(v4)CP believes that the peer PPP device can only have one address.

In IPv6, there's a lot more flexibility regarding prefix utilization
on a host or network.

IPv6 Link-Local addresses have a fixed 64 bit upper portion, and so
the 64 bit token used for IP6CP may be usable to generate a valid
Link-Local address.

Such an address can then be used in DHCP, or Neighbour Discovery.

Since Router Discovery (part of Neighbour Discovery) can provide
multiple prefixes to a host, any one or more of these prefixes
may be used to autoconfigure an address.

As you mentioned, prefix delegation caters to another set of devices:
those requiring networks/prefixes of addresses.

Each fills its own requirement, and it doesn't necessarily make sense
to define another way to do this in IP6CP, since the assumptions
are different to IPv4.

Yours Sincerely,

Greg Daley

Protect your PC! Call in the experts! http://www.msn.co.in/security/ Click here now!

-------------------------------------------------------------------- IETF IPv6 working group mailing list ipv6@xxxxxxxx Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------