marka at isc.org
Mon Mar 4 13:14:42 EST 2013
In message <5133F9A2.4070507 at purdon.id.au>, Bob Purdon writes:
> > Not this old, discredited idea again.
> >> I'm thinking of the universities that are sitting on /16's, yet
> >> announcing only a /24 or two from that range. I believe the US
> >> military, and various large corporations are also guilty of sitting on
> >> /8's.
> > So are you going to pay the costs of renumbering their networks
> > because the addresses are in use whether you can see that they are
> > in use or not?
> > Note just because the addresses are not announce to the public
> > internet doesn't mean that they are not annouced or need to be
> > unique with respect to the public internet. There are networks
> > that see these announcements and also see those on the public
> > internet.
> Isn't this partly what RFC1918 space is for?
Go read RFC 1918. It isn't a panacea. If it works for you well
and good. If it doesn't you shouldn't be forced to use it.
Go think about what happens when you need to connect privately to
thousands of companies using RFC 1918 themselves. At least one of
the companies with a "used /8" is doing just that.
> I just don't see it as appropriate that organisations chew up public
> internet resources (IP addresses), but don't use them on the public
> internet. I'm sure I am not alone in that.
While you are not lone you are misguided.
> RFC1918, as we all know, provides for private addresses for use in
> private networks.
!publically visible != private.
> >> The NIC's should be looking at what is announced and for space that's
> >> not the owners should be required to announce it (with useful/meaningful
> >> services occupying it), or hand it back.
> > So you want the NIC's to break their existing contracts to "reclaim"
> > addresses that may or may not be in use. Are you will to pay their
> > legal fees when it is almost certain that they will loose.
> Do the existing 'contracts' provide for eternal allocation of an unused
> resource? Are they even contracts? My memory isn't good enough to
> remember the correspondence I went through when getting free space from
> AUNIC a bazillion years ago (and I've not been through it with any NIC
> outside of AsiaPac, so I don't know what the rest of the world has done).
Well I remember when I got addresses from SRI. There was no
requirement to connect, in fact you needed a sponsor to connect.
You just needed to require globally unique addresses. The amount
of address space you got was based on your immediate and projected
needs. This was pre CIDR, pre RFC 1918. Most of the networks you
are targeting got their addresses under similar conditions.
> >> If it's not in the global routing table then you don't need it (you
> >> should renumber into RFC1918 space and NAT, since if you are using the
> >> space you're obviously NATing it).
> > Discredited logic.
> I'm not sure your argument above discredited it.
The use or not of RFC 1918 space, once it was created, was up to
the organisation. There was *never* a requirement to use RFC 1918
space if you didn't want to. There still isn't a requirement to
use RFC 1918 if you don't want to.
> You indicated, as I understand it, a need for some organisations to have
> unique IP space to enable communication between themseleves, in addition
> to communication with the public internet.
> Other than those truly massive private inter-networks for which there is
> not enough RFC1918 space, the remainder could use RFC1918 space, no?
Perhaps, perhaps not. That said even if you could recover every address it
is still not worth it.
> Happy to be proven wrong. I've not worked for an organisation that's
> had this problem.
> >> If you are announcing it, but nothing is listening (even residential DSL
> >> users are often pingable and a small percentage have something
> >> listening) - then you're probably using it for outward initiated
> >> connectivity, in which case you can most likely use NAT and don't need
> >> the space either.
> > There are lots of homes that only listen some of the time. With UPnP
> > listening gets turned on and off all the time. The users often don't
> > know that they are listening.
> The point being that in a given /19 for example, there would be enough
> listening to know it's being used. I'm sure there's a way to see if a
> given block of space is likely to be used.
> > The inevitable is here now. The future has arrived. Get over it
> > and deploy IPv6. You will find that most of your equipment already
> > supports it. IPv4 is starting to be come a stinking, rotten corpse.
> I do have IPv6, and have done for quite some time. The last network I
> managed was fully dual-stacked.
Mark Andrews, ISC
1 Seymour St., Dundas Valley, NSW 2117, Australia
PHONE: +61 2 9871 4742 INTERNET: marka at isc.org
More information about the AusNOG