[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [GLOBAL-V6] Comments on AP Consensus
Hi,
On Tue, Apr 16, 2002 at 04:32:27PM +0100, Dave Wilson wrote:
> > I'm not sure whether there is an existing document, but I have seen
> > discussions to this - "just" use multiple /48s, which IPv6 stacks and
> > applications should be prepared for.
> >
> > Similar to what people already do in IPv4 world, except that "many IPs
> > on one single machine for one single service" is something that is
> > natural for IPv6 and unusual for IPv4.
>
> Even if it works, I don't see how it's possible to implement a proper
> routing policy using this method. e.g.: if HEAnet has a routes to JANET
> via (a) high speed direct connection, (b) a high-speed network, and (c)
> an ordinary commodity internet connection, how does the client PC know
> which of the three source address to use?
Good question -> point this to the IETF people (I could imagine a few
variants, which I don't like too much, so I assume that they can come
up with something better).
> If we simply take a /48 from one of the "upstreams" and advertise that
> to all our "upstreams", why should they accept our advertisements?
> Strictly speaking, they shouldn't. (although I dare say if cnn.com did
> so, lots of ISPs would be under pressure to stop filtering when CNN's
> primary upstream goes down).
It might make sense for "your upstream" or "your regional ISPs" to
accept those /48s. I, for example, would want to do that for my
customers.
It does NOT make any sense to make the IPv6 prefix for every single
company visible world-wide. This is what is happening today with
the "BGP multihoming with your own networks will solve all our problems"
paradigm, and this is BAD.
> I'm sorry if I'm rocking the boat, and if urgency is that great for a
> global policy then I guess there's not much can be done.
The problem is that there are people (especially in the APNIC region)
that want to go ahead and actually *use* IPv6. Other people (especially
the ARIN region) don't really see the need for IPv6, but mightily want
to avoid any potential mistakes, so "better no policy than a one that
might turn out to be wrong in 10 years".
Discussion about a new policy (as the old bootstrap phase is over) has
been ongoing since at least half a year, and it is turning in circles,
with the same arguments being rehashed by various people again and
again (well - including me, of course). After the last round in
January/February, people actually made a proposal that got consensus
from the ARIN people, so that would be a good starting point for a new
*interim* policy - to be re-discussed later, of course.
!!! This isn't going to be the final policy !!!
> I just don't
> think the proposal fixes the problems raised at RIPE. (In particular I
> feel that deciding to give any LIR an allocation carries an implicit
> belief that HD-ratio is not a good measurement).
I believe that HD-ratio is a good measurement to figure out whether
an allocation is "full", and a LIR needs "more space".
For "is this LIR worthy to get an allocation", I don't see it as useful
- I said this before: as the kind of customers that a LIR has is so
vastly different, but the IETF rule says "a /48 to each of them",
I see any criteria based on absolute numbers as pretty useless.
Examples are "AOL" (horrendous amounts of /48s to single-user
end-sites - bad usage inside the /48, but lots of /48s used) and a
national research network in a small country, that serves only
"a hand ful" of universities (few /48s, but good usage inside of
those /48).
Not solveable by math.
... but maybe the 200-customer-figure will work. Wilfried should say
something to this, from the view point of a research network...
Gert Doering
-- NetMaster
--
Total number of prefixes smaller than registry allocations: 74810 (74196)
SpaceNet AG Mail: netmaster@Space.Net
Joseph-Dollinger-Bogen 14 Tel : +49-89-32356-0
80807 Muenchen Fax : +49-89-32356-299
-
- This list (global-v6) is handled by majordomo@lists.apnic.net
