IPv6 mistakes, was: Re: Looking for an IPv6 naysayer...

Mark Andrews marka at isc.org
Thu Feb 17 20:30:20 UTC 2011


In message <AANLkTi=UzEQB2DYKxHVrxaKfasPHGfDmXJp1p-GJ0FCf at mail.gmail.com>, Came
ron Byrne writes:
> On Thu, Feb 17, 2011 at 5:08 AM, John Curran <jcurran at istaff.org> wrote:
> > On Feb 17, 2011, at 7:39 AM, Iljitsch van Beijnum wrote:
> >
> >> Not that it matters because it's too late now and it would only give us =
> a few more months, but:
> >>
> >> Does the US government really need more than 150 million addresses, of w=
> hich about half are not publically routed? Non-publically routed addresses =
> can be reused by others as long as the stuff both users connect to doesn't =
> overlap.
> >
> > Again, I note that we've collectively allocated the 95%+ of the address
> > space which was made available outside of DoD's original blocks, and then
> > considering that US DoD additionally returned 2 more /8's for the communi=
> ty
> > (noted here: <http://blog.icann.org/2008/02/recovering-ipv4-address-space=
> />),
> > I believe they've shown significant consideration to the Internet communi=
> ty.
> > The fact that any particular prefix today isn't in your particular routin=
> g
> > table does not imply that global uniqueness isn't desired.
> >
> > Rather than saying 240/4 is unusable for another three years, perhaps the
> > service provider community could make plain that this space needs to be
> > made usable (ala http://tools.ietf.org/html/draft-fuller-240space-02 or
> > http://tools.ietf.org/html/draft-wilson-class-e-00, etc.) on a priority
> > basis and work with the operating system and vendor community actually
> > to make this happen? =A0There's a chance that it could be made usable wit=
> h
> > sufficient focus to make that happen, but it is assured not to be usable
> > if eternally delayed because it is "too hard" to accomplish.
> >
> 
> +1
> 
> If you want to go on a wild goose chase, start chasing down 240/4 and
> you might make some progress.
> 
> As i have mentioned before, it was only after i gave up on 240/4 for
> private network numbering that i really earnestly took on IPv6-only as
> a strategy.  Seeing 240/4 actually work would be nice, but i have
> already concluded it does not fit my exhaustion timeline given how
> many edge devices will never support it.
> 
> If i have to fork lift, it should be for ipv6.

You can reflash CPE devices to support this that you can't reflash
to support IPv6 as there is no space in the flash for the extra
code.  This should be minimal.  A extra PPP/DHCP option and a check
box to enable (default) / disable setting it.

It can be deployed incrementally.

It enables IPv6 to be deployed over intermediate hardware that
doesn't support IPv4.  You still need lots of IPv4 to do that. It
doesn't however have to be globally unique and it shouldn't be RFC
1918.  Leave RFC 1918 for customers.

You add IPv6 support to CPE devices where you can.

It doesn't require the world to upgrade.

It gives a well defined range that you don't use with 6to4.

We also don't need all of class E.  The first half would be more
than enough for even the biggest ISP.

It's big enough to give customers stable IPv6 addresses via 6rd.

Mark

> Cameron
> =3D=3D=3D=3D=3D=3D=3D
> http://groups.google.com/group/tmoipv6beta
> =3D=3D=3D=3D=3D=3D=3D
> 
> > /John
> >
> > (my views alone; 100% recycled electrons used in this message)
> >
> >
> >
> 
-- 
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 NANOG mailing list