Date: Fri, 20 Jul 2001 17:11:35 -0700
From: "Dave Thaler" <[EMAIL PROTECTED]>
Message-ID:
<[EMAIL PROTECTED]>
| First of all, RFC 2373 mandates that all 2000 addreses (and most others)
| are required to have 64-bit interface identifiers,
Not quite, I think what it mandates is that 64 bit interface identifiers
be used for all nets in that space.
It is perfectly OK to redefine smaller ID spaces if you know what you're
doing (and know that this disables autoconf). Eg: it is quite common
to use /127 or /128 prefixes on p2p links.
| so the prefix in question would have to be 2000::/64 (a /96 is illegal).
It makes no difference to the question asked, but I wouldn't say illegal,
just unlikely.
In particular, we do want to make sure that implementations make no
assumptions about prefix length, other than what is strictly required
for correct operation - you need a /64 (or wider) for autoconf, but if
autoconf isn't being used, there should be no limits.
| > Suppose if there are multiple routers attached to that ethernet interface,
| > and a packet comes with a destination address of 2000::/128 which
| > router will respond to that packet.
|
| Technically "any" one of them would be legal, but the answer that makes
| the most sense is "the first one that the packet reaches" (i.e. the one
| closest to the source).
If the source is off the net, that is the way it is. I kind of suspect
though that the question more assumed that the source was on the 2000::/96
(or 2000::/64 ...) ethernet.
In that case the answer is "probably all of them". And typically the
first response that arrives will be used, and the others will be discarded.
kre
--------------------------------------------------------------------
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]
--------------------------------------------------------------------
