[6bone] semi-newbie Q on IPv6 address planning

[Pekka Savola]

On Fri, 2 Aug 2002, Gert Doering wrote:
> > http://search.ietf.org/internet-drafts/draft-savola-ipv6-127-prefixlen-0
> > 4.txt
> 
> Done.
> 
> I don't agree with many of the conclusions - why would it be "the best"
> solution to assign /64s to ptp links?  There is nothing "good" about /64s.

The "good" about /64's is that this is what protocol's address 
architecture requires.  That's something that future extensions (e.g. 
address based keying for IPSEC) may require.  Therefore it should be the 
first recommendation.
 
> If /127s are not used, what is the benefit of a /112?  Who needs 64k 
> addresses on a ptp link?

To cope with the fact that 128 values have been reserved for anycast (see 
RFC2526), we'd "need" at least /120.  /112 is on the hex boundary so the 
numbers after the last ":" are always the nodes.  There is enough space in 
a /64 to do enough /112's.  It just seems best, if you don't want to use 
e.g. /126 or /64.
 
> The general issue of anycast addresses on ptp links doesn't sound very
> useful either, but this seems to be an unavoidable issue - so thanks 
> for pointing it out, we'll consider moving to /126s.

Once it gets implemented (and especially if it's automatically enabled), 
it could very hairy..

-- 
Pekka Savola                 "Tell me of difficulties surmounted,
Netcore Oy                   not those you stumble over and fall"
Systems. Networks. Security.  -- Robert Jordan: A Crown of Swords