[arin-ppml] IPv6 Heretic thoughts

[John Santos]

On Fri, 5 Sep 2008, Dan White wrote:

> John Santos wrote:
> > On Fri, 5 Sep 2008, Dan White wrote:
> >   
> >> I think there's an implication in this line of reasoning that IPv6 is 
> >> hard, and that dual stack is bad.
> >>     
> > I think there are 2 serious issues with this.  1) It doesn't immediately
> > get you anything (with IPv4 runout) because each dual stacked host
> > needs *TWO* addresses, an "easily obtainable" IPv6 address and a scarce
> > IPv4 address.  
> 
> I don't grasp your point. Are you saying that it's a management problem 
> of maintaining two addresses? Arguably, there is no end-user maintenance 
> of IPv6 addresses (in a dual stack scenario).

No!  Go back to basics.  What problem are we trying to solve?
The problem is *not* getting people to adopt IPv6.  The problem is
running out of IPv4 addresses.  IPv6 is a potential *solution* to this
problem, and any difficulties getting people to adopt IPv6 are
secondary problems, not the primary problem.  Dual-stacking does
*not* solve the primary problem, so as a method of IPv6 adoption, it
is only useful if it is temporary or limited in scope.  Universal
dual-stack is not limited in scope, so it must be limited in time or
it doesn't solve the problem.  (That said, I think dual-stacking is
necessary for limited scope (i.e. servers), and might be a good 1st
step toward IPv6 adoption, but it has to be achieved *before* IPv4
runout, or we still hit the wall.  And once it has been acheived,
it is no longer necessary!  Because if everything accessible via
IPv4 is also accessible via IPv6, IPv4 can be shut off with no loss
of connectivity.)

> 
> > 2) Due to the way IPv6 defaults (apparently this is
> > defined in the specifications), it tries an IPv6 connection 1st and
> > then fails back to an IPv4 connection.
> 
> Not exactly. In the case where a user is visiting a website, the DNS 
> resolver will do *one* search for IPv4 and IPv6 addresses for that site 
> name. If there are no AAAA records for the site, no harm no foul. If an 
> admin is advertising IPv6 services for a hostname that is not 
> connectible that's a management issue, not a protocol deficiency.

The end site might have a perfectly configured IPv6 network and it's
upstream may be perfectly configured, and so might my network and my
upstream, but if something in between is *not* IPv6, I'll still
encounter this problem.

> 
> > So it's not that dual-stacking is *bad*, but it just isn't too
> > helpful until the infrastructure is up to snuff, and you can start
> > introducing IPv6-only hosts.  But to support existing IPv4-only
> > hosts, *all* publicly accessible servers will have to be dual-stacked
> > for the forseeable future.
> 
> Dual stacking solves the problem of having IPv4 only hosts. It also 
> conveniently solves the problem of having IPv6 only hosts. I do 
> understand that there's a bit of a chicken and egg problem, but if you 
> start dual stacking your hosts, you're at least prepared for what's coming.
> 
> - Dan
> 
> 

-- 
John Santos
Evans Griffiths & Hart, Inc.
781-861-0670 ext 539