Re: [5gangip] New Version Notification for draft-vonhugo-5gangip-ip-issues-01.txt

[Tom Herbert <[e-mail]>]

On Wed, Nov 16, 2016 at 9:04 AM, Dino Farinacci <[e-mail]> wrote:

> Let’s see what the carriers would say about this. And what the 3GPP
> requirements are.
>
> And let's see the code also :-)

Tom


> Dino
>
> > On Nov 15, 2016, at 3:55 PM, Ca By <[e-mail]> wrote:
> >
> > Dino,
> >
> > Yes, ipv6-only UEs are practical. We launched them in android 3+ years
> ago (464xlat).
> >
> > We have ipv6-only iPhone in beta on 10.2.
> >
> > In 2017, T-Mobile US shops simply won't have any phones in the shops
> that have ipv4 addresses assigned from the network or emit native ipv4
> packets across the radio by default.
> >
> > I do not consider that "product suicide".  It is executing a forward
> looking strategy.
> >
> > Regards,
> > Cameron
> >
> > On Tue, Nov 15, 2016 at 3:43 PM Dino Farinacci <[e-mail]>
> wrote:
> > > Sorry for top posting.... gmail broke inline editing on iOS....
> > >
> > > Why is ipv4 required?  I stopped assigning ipv4 to android phones in
> 4.3 and the same will happen to Apple devices in 10.2.
> >
> > I have my reasons below. There will still be services and applications
> that require IPv4. Do you think an IPv6-only UE is practical?
> >
> > The requirements in the presentation yesterday also indicated IPv4
> support.
> >
> > > There is no good reason to support ipv4 (uniquely assigned to each ue)
> in 5G IMHO. Obviously, there needs to be NAT64 to access the legacy
> internet, but that is ipv6 across the 5G RAN and core.
> >
> > We can argue over the requirements but IPv4 is not the only reason to
> have an anchor point.
> >
> > > And, the IAB has stated ipv4 is not required going forward , please see
> > >
> > > https://www.iab.org/documents/correspondence-reports-
> documents/2016-2/iab-statement-on-ipv6/
> >
> > And do you see the vendors rushing to remove it from the stack? That
> would be product suicide.
> >
> > Dino
> >
> > >
> > > Regards,
> > > Cameron
> > >
> > >
> > > On Tue, Nov 15, 2016 at 3:02 PM Dino Farinacci <[e-mail]>
> wrote:
> > > > Very good point. The ILNP opportunity i see is to eliminate the
> middle box.  As where ILA (and LISP) requires the middle box to be the
> anchor point for mobility, right?  These anchor points are frequently
> magnets for operational failure as well as surveillance.
> > >
> > > Both LISP and ILA can run in the host where there are no anchor
> points. But as Tom mentioned, this may be a much harder deployment
> challenge then making a solution network based.
> > >
> > > You are going to have to put anchor points in the network for scale,
> fast handoff, and NAT-traversal. If there is a requirement for IPv4, which
> there should be, NATs will be in the carrier path. In addition, you will
> need anchor points for multicast replication unless the carrier is
> sincerely committed to put native multicast routing throughout the network.
> > >
> > > Those are the reality challenges we face. You need a robust solution,
> and better to be a single solution to reduce complexity, that has all the
> features above.
> > >
> > > LISP is the only protocol that exists (with 10 years of development
> expeience and 6 years of deployment experience) that can do mobility,
> multi-homing, multicast, and multiple address-families at the same time.
> > >
> > > Dino
> > >
> >
>
>