Re: I-D ACTION:draft-blanchet-v6ops-routing-guidelines-00.txt

[Marc Blanchet <marc.blanchet@hexago.com>]

Le 05-09-19 à 02:10, Pekka Savola a écrit :

> Note: the next rev of this doc (if there is one) should maybe be  
> forwarded for comments on the ipv6-ops@lists.cluenet.de list.

will do.

> On Sat, 17 Sep 2005, Durand, Alain wrote:
>
> > How does this differ from RFC2772? Do you feel a need to update it?
>
> RFC2772 includes guidelines for *6bone*.  This is offering  
> guidelines for more than that.  In that light, if we feel the IETF  
> is the right place to give guidance on this, updating is certainly  
> needed.

see comment below.

> As already noted in the IETF/ipv6 lists, it's an open question to  
> what degree and how strong guidance the IETF should (or even could)  
> give to the operators.
>
> It seems that if the IETF would want to pursue this work, the  
> discussion would need to be recast in the manner "You should  
> considering doing [this].  This is why it may or may not make  
> sense". As we learned from the DNS configuration debates at DNSOP  
> (though the situation is a bit difficult), getting to a consensus  
> could be very difficult.
>
> A couple of specific comments:
>
>  - the global unicast address recommendations are (IMHO) way too  
> lax. Even some part of 'strict' filters in http://www.space.net/ 
> ~gert/RIPE/ipv6-filters.html could be stricter. In any case, that  
> URL should also be referred.  This is an example of the case why  
> getting CONSENSUS on a document like this is going to be very  
> difficult.

I knew that... ;-))

well. the intent here was to set the minimum (/48 or less) so that at  
least this is clear and I'm hoping there is a concensus on /48 or  
less. In this case, the document with the other guidelines is useful  
for operators.
for example, if the minimum is /48 or less, then we are filtering / 
64! good progress!

By the time there is concensus on less than /48 (maybe after  
multihoming work or ...), then we could revise it and make a new  
version.

currently (to also respond to Alain comment on diff with RFC2772),  
RFC2772 is the only "published" document I know that has some  
guidelines on routing policies. and RFC2772 is outdated. So there is  
a need for a document, with the minimum that has concensus to help  
the ones who are deploying right now.

> Many folks don't see a problem w/ putting /48 junk etc. in the  
> routing tables, others do.  This doc would/could then become a  
> political battleground in the v6 operations community..

Why don't publish what is agreeable as a minimum and then revise it  
when more experience/more concensus/other tools/... is in.

>  - the uppercase keywords are inappropriate and should be removed

you meant because it is not a protocol?

>  - the doc should probably make RFC2772 historic

I thought about that and I agree. I'm not sure when/how to put it.  
Include in the document I guess?

Marc.


> --
> Pekka Savola                 "You each name yourselves king, yet the
> Netcore Oy                    kingdom bleeds."
> Systems. Networks. Security. -- George R.R. Martin: A Clash of Kings