[AusNOG] My Predictions for the ISP Industry

[Mark Andrews]

In message <4F5FBBE9.8050508 at layer10.com.au>, Paul Brooks writes:
> On 14/03/2012 2:52 AM, Mark Delany wrote:
> >> But, it will take most ISPs a year or two to fully integrate IPv6 into
> >> their networks, and those who haven't started doing it yet, might as well
> >> be planning to shut down their businesses because in the next year or two,
> >> it will be too late...
> > Really? In which decade do you think that facebook.com, yahoo.com,
> > twitter.com and google.com will stop returning an A RR? And what is
> > their incentive for doing so?
> 
> Its not a question of stopping returning an A RR.
> Its a question of when they'll start to return AAAA RRs, when queried for one
> .
> Especially when the AAAA query is within an IPv6 packet.
> The answer is June 6th this year - http://www.worldipv6launch.org/
> 
> Dual-stack hosts query for a AAAA record first, and then fall back to queryin
> g for an
> A record if they don't get a reply.

If they don't get reply then there is a broken nameserver that
doesn't meet STD 13 requirements.  With properly functioning
nameserver you should ALWAYS get a reply even it it says "there are
no AAAA records for this name".

That said there are still broken load balancers out there that don't
reply or return NXDOMAIN / NOTIMP / SERVFAIL for AAAA queries
whichand misconfigured load balancers out there that return the
*wrong* SOA record as the load balance has been configure as if it
has been delegated example.com instead of wwww.example.com.

If your organisation has one of these then you should fix it or
replace it.  DNS isn't rocket science.  It isn't hard to check that
it returns the correct answer for ALL query types.  There are only
65536 of them which takes a couple of seconds to check.

> The trick is to make sure they don't rece
> ive a
> query for an A RR in the first place, because if they do - the customers ISP 
> has failed.

No.  There will always be a A queries.  The traffic however should
shift to being mainly IPv6 with IPv4 on error unless the OS vendors
remove the slight IPv6 bias like Apple did with Lion.

> Its not Facebook and Yahoo that have failed here - its Bigpond, Optusnet, and
>  all the
> other ISPs that aren't yet handing out IPv6 addresses in their DHCP responses
> .

Or telling their customers what they need to buy so they can connect
over IPv6 when they turn it on.   Add to that the CPE router vendors
that don't have IPv6 support in low end boxes yet.  It is appearing in
high end boxes.

> > You need to explain why a business would voluntarily stop listening to
> > IPv4 traffic and why ISPs would stop carrying it.
> Mark - this is all wrong. Nobody has to stop listening or carrying IPv4. They
>  just
> have to start listening to and carrying IPv6 as well.
> 
> 
> >
> > If you have no explanation for that, then what has any ISP got to lose
> > by just carrying IPv4? After all, it gets to everywhere and probably
> > will do so for a very long time into the future.
> >
> > The big problem is that turning off IPv4 has no value-add and turning
> > on IPv6 has no value-add, so no one cares to do either. Ergo,
> > IPv4-only systems will continue to work for the foreseeable
> > future. That means there is zero imperative to support IPv6.
> >
> > I'll be proved wrong when any major website discards their A RR web
> > site and only advertise an AAAA web-site. Any volunteers? Apnic? Arin?
> > ICANN? eintellego.net? Jut curious. If IPv6 is such a hit, when do you
> > guys plan to drop your IPv4 RR?
> >
> >
> > Mark.
> > _______________________________________________
> > AusNOG mailing list
> > AusNOG at lists.ausnog.net
> > http://lists.ausnog.net/mailman/listinfo/ausnog
> 
> _______________________________________________
> AusNOG mailing list
> AusNOG at lists.ausnog.net
> http://lists.ausnog.net/mailman/listinfo/ausnog
-- 
Mark Andrews, ISC
1 Seymour St., Dundas Valley, NSW 2117, Australia
PHONE: +61 2 9871 4742                 INTERNET: marka at isc.org