Re: [dhcwg] Prefix-length of an assigned address, in a DHCPv6 message

["Hemant Singh (shemant)" <[e-mail]>]

Few years back, in the 6man mailer, we have discussed this issue.  If the router has different prefix length than what the DHCPv6 server assigned to the client, who wins?

Hemant

-----Original Message-----
From: dhcwg [mailto:[e-mail]] On Behalf Of Bernie Volz (volz)
Sent: Tuesday, September 09, 2014 11:28 PM
To: Ralph Droms; Simon Hobson
Cc: dhcwg
Subject: Re: [dhcwg] Prefix-length of an assigned address, in a DHCPv6 message

Perhaps it has been raised already in this thread, but there is again a draft on using DHCPv6 for some ND aspects:

http://datatracker.ietf.org/doc/draft-sarikaya-dhc-dhcpv6-raoptions-sadr

I am not advocating this work, just pointing out it is yet another attempt to distribute prefix information via DHCPv6.

There already has been an email discussion with the author that this work is probably best done in another WG (if at all).

- Bernie

-----Original Message-----
From: dhcwg [mailto:[e-mail]] On Behalf Of Ralph Droms
Sent: Tuesday, September 09, 2014 4:35 PM
To: Simon Hobson
Cc: dhcwg
Subject: Re: [dhcwg] Prefix-length of an assigned address, in a DHCPv6 message

Simon ...

On Sep 9, 2014, at 12:14 PM 9/9/14, Simon Hobson <[e-mail]> wrote:

> Perhaps it's just me struggling to break out from "IPv4 thinking", but I struggle to see what use an IPv6 address is without explicitly or implicitly a prefix length to go with it. Unless you restrict yourself to "on link" communications, at some point you need to work out whether, when using that address, the other address you wish to communicate with is local (on link) or remote.

The way I think about IPv6 addresses, prefixes and prefix length is that the prefix length is associated with the prefix, not the address.  In typical usage, the prefix is matched against the leftmost bits of the address corresponding to the prefix length and a match indicates the address is in the subnet defined by the prefix.

In my opinion, the "IPv4 thinking" is the association of a prefix length with an address, implicitly defining a prefix; for example, an on-link prefix derived from an assigned address.  In IPv6, prefixes and addresses are more clearly delineated and the prefix lengths are associated explicitly with the prefixes.

> 
> I know there's the arguments that in many large organisations, the routers and DHCP servers are managed by different groups - but the groups still need to agree on what the prefixes and prefix lengths are for any network !

Yes, there need to be agreement - and the agreed upon conventions are typically expressed in ND PIOs, which give the lengths for the advertised prefixes.

- Ralph

> 
> _______________________________________________
> dhcwg mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/dhcwg

_______________________________________________
dhcwg mailing list
[e-mail]
https://www.ietf.org/mailman/listinfo/dhcwg

_______________________________________________
dhcwg mailing list
[e-mail]
https://www.ietf.org/mailman/listinfo/dhcwg