Re: [v6ops] DHCPv6 vs ND strikes again

[Mark Smith <[e-mail]>]

On Sat, 23 Oct 2010 00:26:00 +0200
Ole Troan <[e-mail]> wrote:

> Brian,
> 
> >>> To add to Tony's point, I have to agree that there are folks that do not
> >>> have experience operating networks, specifically deploying IPv6, that seem
> >>> to enjoy asserting their assumptions and views.  Instead of doing this it
> >>> might be useful (and more productive) if folks first tried to understand how
> >>> various types of networks work before making recommendations, etc.
> >> 
> >> perhaps it would be equally productive if the operators tried to understand how end-users would want to build their networks?
> >> 
> >> I base my previous assumption on RFC3177. as we're on an IETF list, is there any other IETF recommendation for a different allocation model?
> > 
> > On 2010-10-18 07:00, Fred Baker wrote:
> > 
> >>>> Next week (I'm giving you an extra week's notice as I know you are busy), the IETF IPv6 Operations Working Group is initiating a two week working group last call of 
> >>>> 
> >>>> http://tools.ietf.org/html/draft-ietf-v6ops-3177bis-end-sites
> >>>>  "IPv6 Address Assignment to End Sites", Thomas Narten, Geoff Huston,
> >>>>  Rosalea Roberts
> 
> do you think rfc3177bis changes this?
> i.e. does it no longer advocate that "sites should get sufficient address space without jumping through hoops"? and that this should be a long lived address block assignment, not one which is subject to dynamically change as the end user network grows or shrinks?
> 
> I have no issue with delegation of address space of arbitrary length.
> I have an issue with:
>  - the expectation that an RR will know what to ask for
>  - that RFC3633 provides a suitable mechanism for negotiating address space size
> 
> as one of the authors of RFC3633, I think it is fair that I declare what assumptions that document was written under. my assumption was one of delegation of long-lived prefixes without any negotiation, i.e. with assignment size set by the delegating routers. and that is in the spirit the document is written. if operators want to delegate prefixes in a different fashion, they have to be aware that the tool (3633) may not be up for the task.
> 

If you start trying to play the 'right sizing' game, you're
treating the IPv6 address space like it is scarce. You're unnecessarily
increasing your operational effort because you'll now have to monitor
prefix utilisation, periodically add to the pool, make projections
about future utilisation, possibly deal with a service outage if you
don't make good future projections and exhaust the pool (likely at
night in a residential environment), and record more details about who
had what prefix at what time and now also the prefix length that they
had at that time.

Or you can just throw address space at the problem, giving
most typical customers more than they're likely to need, and
eliminate all this unnecessary effort and risk.

Regards,
Mark.