>>>>> On Sun, 9 Dec 2001 21:17:28 -0800,
>>>>> "Richard Draves" <[EMAIL PROTECTED]> said:
>> - The longest matching algorithm is not so meaningful in the
>> destination address ordering.
> Well, I think it is quite meaningful. It can certainly go awry (per
> your earlier example), but I think more commonly it will be
> helpful. For example, if node A is part of a multihomed site and so
> has two global address P::a and Q::a (from ISP's P and Q) and node B
> is connected via ISP P and so has the global address P::b, then when
> node B connects to node A the longest-matching prefix heuristic in
> the destination address ordering will cause node B to first try P::a
> and then Q::a.
Ah, yes, you're right. I just missed the symmetric story of the
source address selection among multiple candidates.
So, IMO, the point is
do we need the explicit longest matching while the generic policy
table can do this and the effective prefix length can vary even
today?
I myself do not have a strong opinion on this, but if I need to vote,
I'll be for leaving the current spec.
JINMEI, Tatuya
Communication Platform Lab.
Corporate R&D Center, Toshiba Corp.
[EMAIL PROTECTED]
--------------------------------------------------------------------
IETF IPng Working Group Mailing List
IPng Home Page: http://playground.sun.com/ipng
FTP archive: ftp://playground.sun.com/pub/ipng
Direct all administrative requests to [EMAIL PROTECTED]
--------------------------------------------------------------------
