[e2e] Discrete IP

[Daniel Havey]

a simple soho-box which may well play
> around with NAT and congestion management and queue
> management and all these funny little things which PhD
> students change the world with, without being noticed by the
> latter.

Hey Detlaf ;^)  How did you guess what I have been doing over the summer?

> > What do you mean by "parallel"? Why do you assume that
> the protocols will be run in "parallel?"

I got this one.  At the core routers the different IPvXs would have to run in parallel.  The header format is different because of address lengths etc.  The router must examine the version number and then act appropriately.  The router must understand the header format of all packets that it will forward.

You have a chicken, and you have an egg.  Why would a company like Cisco build routers that understand IPv7 - IPv42?  They built IPv6 in because they had a solid business reason.  There weren't enough IPv4 addresses.  IPv6 = more customers.  So they built it and then the customers bought it.

Since there are plenty of IPv6 addresses, all that is left to convince the core to change is some random "feature set".  Not a very strong business reason.  They will not build it.

You can change the world, but you cannot change a core router ;^)

...Daniel (PhD student, who is busy changing the world ;^)


--- On Sat, 9/15/12, Detlef Bosau <detlef.bosau at web.de> wrote:

> From: Detlef Bosau <detlef.bosau at web.de>
> Subject: Re: [e2e] Discrete IP
> To: 
> Cc: end2end-interest at postel.org
> Date: Saturday, September 15, 2012, 6:20 AM
> On 09/15/2012 06:52 AM, Pars Mutaf
> wrote:
> > Hi Detlef,
> > 
> > On Sat, Sep 15, 2012 at 2:35 AM, Detlef Bosau <detlef.bosau at web.de
> <mailto:detlef.bosau at web.de>>
> wrote:
> > 
> >     On 09/13/2012 06:38 AM, Pars
> Mutaf wrote:
> > 
> >         Hi Andrew,
> > 
> >         China has IPv6
> for example but I cannot talk to them.
> >         I don't have to
> install IPv6 to talk to them. If one day
> >         someone uses IPv7
> (it is their right), I don't have to install
> >         IPv7.
> >         The fact that
> there is a version field doesn't mean that all
> >         versions are
> supported.
> > 
> > 
> >     I think, you miss two basic
> points.
> > 
> >     First: There is no such thing
> as "the" end to end principle.
> >     Particularly, Internet
> communication is nothing which happens
> >     between the communication end
> points and only there, but most of
> >     the work is done at the nodes
> in between.
> > 
> > 
> > 
> > This contradicts what you say below. Below you say that
> you want to avoid this complexity. Here you say that this
> the way it is. So I don't understand you message.
> > 
> 
> 
> I don't see a contradiction here. Avoiding complexity does
> not mean that complexity does not exist.
> 
> End nodes hardly have to deal with packets. Every now and
> then, there is a packet to be sent or to be received. So,
> end nodes have any time they want to inspect packets, to
> interpret them, whatever they want. Routers in the middle
> miss this privilege. They may be offered up to millions of
> packets each and every second. So, the effort spent for
> serving a single packet must be kept as small as possible.
> 
> And of course, there is a huge difference between a core
> router in the tier 1 backbone, which has to deal with huge
> amounts of data, and a simple soho-box which may well play
> around with NAT and congestion management and queue
> management and all these funny little things which PhD
> students change the world with, without being noticed by the
> latter.
> 
> Hence, although the IETF cannot make the world run IPv6, our
> common interest is to switch over to one common protocol in
> the internet. At least for the tier 1 backbone or other
> extremely busy parts of the Internet.
> > 
> > 
> >     Second: The Internet is an
> overlay network by design. We want ONE
> >     common protocol which is
> supported by all nodes connected to this
> >     overlay network. Particularly,
> it shall not be the intention of
> >     the Internet to run several
> protocols in parallel. Nevertheless,
> >     this happened in the past,
> happens in the present and is expected
> >     to happen in the future,
> however it is not the basic intention.
> >     The more protocols you run in
> parallel, the more complex your
> >     intermediate nodes, which do
> all the routing work, will be. And
> >     it's certainly not our goal to
> make thinks unnecessarily complex.
> > 
> > 
> > What do you mean by "parallel"? Why do you assume that
> the protocols will be run in "parallel?"
> > 
> > 
> > 
> > 
> > 
> >     DB
> > 
> > 
> >     -- 
>    ------------------------------------------------------------------
> >     Detlef Bosau
> >     Galileistraße 30
> >     70565 Stuttgart   
>                
>         Tel.: +49 711 5208031
> > 
>    <tel:%2B49%20711%205208031>
> >               
>                
>                
> mobile: +49 172 6819937
> > 
>    <tel:%2B49%20172%206819937>
> >               
>                
>                
> skype: detlef.bosau
> >               
>                
>                
> ICQ:  566129673
> >     detlef.bosau at web.de
> <mailto:detlef.bosau at web.de>
> >     http://www.detlef-bosau.de
> > 
>    ------------------------------------------------------------------
> > 
> > 
> > 
> > 
> > -- http://www.content-based-science.org
> > 
> 
> 
> --
> ------------------------------------------------------------------
> Detlef Bosau
> Galileistraße 30    
> 70565 Stuttgart           
>                
> Tel.:   +49 711 5208031
>                
>                
>            mobile: +49
> 172 6819937
>                
>                
>            skype: 
>    detlef.bosau
>                
>                
>            ICQ: 
>         566129673
> detlef.bosau at web.de 
>                
>    http://www.detlef-bosau.de
> ------------------------------------------------------------------
> 
>