Re: [v6ops] Operational Implications of IPv6 Packets with Extension Headers - implications from new development for EHs

[Tom Herbert <[e-mail]>]

On Wed, Jul 29, 2020 at 8:35 AM Gert Doering <[e-mail]> wrote:
>
> Hi,
>
> On Wed, Jul 29, 2020 at 08:18:28AM -0700, Tom Herbert wrote:
> > > There is no incentive for network operators to make sure their network
> > > can transport EHs if it works "without".
> > >
> > > There is no incentive for application developers to code support for
> > > "anything with EH" if they have to assume "it will need to work across
> > > networks without".
> >
> > Then they won't be able to use segment routing, IOAM, or new QoS
> > signaling like Network Tokens that are now being developed and being
> > deployed. That's fine, it's their prerogative.
>
> And by that, you acknowledge that using "EH across the wide Internet"
> is not going to work.
>
Gert,

Well, if by "works across the wide Internet" you mean that there is a
100% probability that packets with extension headers will reach every
possible destination over every possible path then, yes, "EH across
the wide Internet" is not going to work. However, by this same
definition IPv6, UDP, SCTP, IP fragmentation, ICMP, and pretty much
anything except possibly the most rudimentary plain TCP/IPv4 packet
don't work "across the wide Internet" either. We know ubiquitous
support for protocols across the whole Internet is not feasible, it's
pointless to assume a requirement for a protocol that it must be
universally supported everywhere before it can be useful.

> Nobody challenged that the use of EH in a closely controlled environment
> ("my network and all the partner networks that I have a contract that
> governs that we'll cooperate regarding these EHs") will work.
>
> [..]
> > but IMO if we apply the right techniques we can opportunistically
> > provide something that works and provides value to users.
>
> EH has turned out to be costly to people that have no way to recover
> these costs as their customers have no interst.
>

I don't expect customers to have any interest in extensions headers, I
do expect they'll have interest in services they help provide.

> Now, if Google would show up and say "all web sites that demonstrate
> use of the TurboEH are ranked up by 20%", you can bet that SEOs would
> be the first to demand working and fully transparent EH support, no
> matter the cost...

Yep. This was clearly demonstrated by Google's success in pushing
operators to support QUIC. If I was endeavouring to bring up a new
protocol or extension header on the Internet I'd certainly start with
getting a major content provider on board even before service
providers.

Tom



>
> Gert Doering
>         -- NetMaster
> --
> have you enabled IPv6 on something today...?
>
> SpaceNet AG                      Vorstand: Sebastian v. Bomhard, Michael Emmer
> Joseph-Dollinger-Bogen 14        Aufsichtsratsvors.: A. Grundner-Culemann
> D-80807 Muenchen                 HRB: 136055 (AG Muenchen)
> Tel: [phone]         USt-IdNr.: DE813185279