Re: Catalog of IPv4 literals

[bmanning@vacation.karoshi.com]

On Sun, Jul 25, 2010 at 02:02:59PM -0700, Cameron Byrne wrote:
> On Sun, Jul 25, 2010 at 1:08 PM,  <bmanning@vacation.karoshi.com> wrote:
> >
> > if an application "breaks" becuase someone uses an address literal instead
> > of a domain name, then that application is itself broken.  the DNS translates
> > the name into an address and the address is used...  so whether a name or
> > a literal is handed to the app should be immaterial.
> >
> 
> Please think of the scope as limited to
> draft-ietf-behave-v6v4-framework scenario #1, IPv6 network to an IPv4
> internet.  If an IPv4 literal is passed at the application layer
> (HTML, XML, ...) to a host with only IPv6 connectivity, the service is
> broken to an IPv6-only user as where it works for an IPv4-only user.
> Pedantry aside, this is the customer experience for some common
> internet services.  DNS names solve this problem since they allow
> DNS64 to function.

	ok... then its a self-inflicted wound. IPv6 mapped addresses
	worked just fine until this wg depricated them.
	wrt DNS64,  the translate functions I've been using for several
	years now doesn't require such hacks. 

> This  draft draft-wing-behave-http-ip-address-literals attempts to
> work around the problem of IPv4 literals passed to hosts in IPv6-only
> networks, but the work around is only relevant for HTTP and will not
> work for smartphones or dumbphones that don't have this proxy logic.

	yes, this turns out to be (as previously noted) an APPLICATION
	level problem.  and i really question the wisdom of spending too 
	much time on this corner/transition case.

> > so your "catalog" of address literals is really the full set of all IP addresses.
> >
> 
> No.

	if you place all of the above restrictions on your premise, then I
	agree with you.

--bill
> 
> Cameron
> 
> > --bill
> >
> >
> > On Sun, Jul 25, 2010 at 09:01:12AM -0700, Cameron Byrne wrote:
> >> Folks,
> >>
> >> It has been suggest several times to me that IPv4 literals be
> >> cataloged in a central location so that those working to develop
> >> IPv6-only networks and services can know the impact of IPv4 addresses
> >> that are hard-coded into content and protocols.  So, i created this
> >> Google Groups http://groups.google.com/group/ipv4literals and provided
> >> an example template for reporting found IPv4 literals.  Right now, the
> >> threat of IPv4 literals on IPv6-only networks is small from the
> >> network operator perspective, it is not a blocking issue.  But, for
> >> the content owners who knowingly or unknowingly have IPv4 literals as
> >> part of their service, this is major breakage.  That said, they have a
> >> right to know how their service will break so that they can accept the
> >> risk of having their content unavailable on major networks or work to
> >> use DNS names that will function correctly.  Extra bonus points if
> >> they resolve this issue of inter-operating with IPv6-only networks by
> >> producing native IPv6 content!
> >>
> >> In my own efforts, i have found content owners very happy to receive
> >> this proactive notification.  Explicitly, myspace and Yahoo! have been
> >> very good partners in finding and resolving issues of this nature and
> >> removing IPv4 literals from their production services.  Also, over the
> >> course of my work I have seen Hulu.com independently move to using DNS
> >> names.  The issue is most commonly found with streaming services on
> >> the Internet, especially ones involving CDNs.
> >>
> >>
> >> Best regards,
> >>
> >> Cameron
> >