Date: Tue, 19 Dec 2000 20:37:10 -0800
From: Richard Draves <[EMAIL PROTECTED]>
Message-ID:
<[EMAIL PROTECTED]>
| You get into these issues in more detail below, but the quick summary is
| that applications can select their own addresses (by binding to a specific
| source address) and the draft recommends but does not require the strong
| host model, per WG consensus.
Would it not be nicer for an application that wants the strong model to
bind to an interface, rather than (or in addition do) binding to an address?
All kinds of strong model problems (like handling separate transmit/receive
interfaces - the application would bind to the receive interface, and
dealing with multiple interfaces with the same local address) get sorted
out comparatively easily that way. All it leaves open is the method by
which an application binds to an interface.
Note, this question is directed more to the WG in general than to the
draft authors.
Also note that perhaps this reflects my general preference against using
the strong host model for much at all.
| When routers are forwarding packets, the draft doesn't apply since they are
| not performing source or destination address selection. But when routers do
| perform source or destination address selection, then I believe the draft
| should apply. So I think it is appropriate for the scope of the draft to be
| all IPv6 nodes.
yes, though when operating in that mode it is also possible to consider that
the "router" is really a "host".
| Steve Deering (I think it was) said that some multicast routing protocols
| break if packets are sent with a source address that is assigned to a
| different interface. (I don't know enough about multicast routing to vouch
| for this myself.) Hence this requirement for multicast destinations.
Yes, multicast reverse path routing has some strange effects in the
unusual cases - strong model sending is almost certainly a requirement
there (unless a different modem of multicast routing is invented sometime).
| One of the consequences is that Redirects will not work properly if a host
| uses the weak model.
If it uses the weak model, and the address chosen doesn't match what
would have been used for the strong model.
Sometime or other someone probably needs to do some more work on
redirect - the redirect mechanisms, and rules, basically haven't
changed for 20 years now. There ought be a reasonable way that we
can have redirect and have the weak model (and perhaps even have
redirect to another interface). But all that is for sometime
far far away.
| > the network layer should go ahead and use that source
| > address, even if the
| > routing table happens to dictate that the packet gets
| > transmitted out a
| > different interface.
| Again, this is the same strong vs weak argument.
Yes, though in this case (with the strong model) the address selection
should win, rather than the routing selection, and the packet should
continue to get sent via the "wrong" interface. That might mean
it can no longer be delivered of course. But a host "routing" system
is rarely going to be complete enough to know this for sure without
actually trying it.
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]
--------------------------------------------------------------------
