Keith Moore wrote:
>
> > > There is also a potential scaling issue of using multiple addresses
> > > as general purpose multihomging mechanism. This is because if this
> > > is the case, most of the Internet hosts will end up with multiple
> > > addresses.
> >
> > I don't see why this is inherently a problem.
>
> it's a problem because it's essentially asking the sending host to do
> routing in the absence of any routing information.
First thing that came to my mind was "great, looks like token-ring
source-routing." I'd hoped we learned our lesson on that one. Having end
nodes attempt to make network path decisions is a monumentally BAD idea.
>
> if multiple addresses are available for a host, the chances are good
> that the paths associated with some of those addresses are significantly
> better, or worse, than others. with IPv4 multihoming, the routing system
> sorts out which path to use. this doesn't work perfectly but at least
> the decision is made in light of some information about the nature and
> current state of those paths. with IPv6 multihoming, the sending host is
> just guessing. it's difficult to believe that this will work well.
I can imagine that a company which is multihomed to two or more ISPs,
each of whom is tier-2 or below, may well have a dozen addresses to deal
with, per host. Now somebody remind me of why we wanted all this extra
address space? It was so we could give every machine a dozen addresses?
Exactly how the end nodes are to know which of these addresses to use,
especially when the decision point in the topology is several layers
above (at the tier2 to tier1 provider attachment point) is going to be
interesting.
Token-ring source routing allowed relatively dumb bridges to be used.
This protected the "core" of the network from having to have extra
processing horsepower. Of course we soon saw Ethernet bridges and
switches which could handle all of the needed decisions without the
involvement of the end stations. We started down the path of MPLS to
protect the core routers which were going to melt under the traffic
load. Then vendors showed us hardware which could route packets at wire
speed. Now we're busy pushing technology to ensure routing table sizes
won't increase because the present generation of hardware can't handle
larger routing tables. Are we being a bit short-sighted?
--
-----------------------------------------------------------------
Daniel Senie [EMAIL PROTECTED]
Amaranth Networks Inc. http://www.amaranthnetworks.com
