[e2e] Free Internet & IPv6

Neil Davies neil.davies at pnsol.com
Wed Sep 19 02:40:24 PDT 2012

Ah Jon, if only it was that easy.

We've been looking at this from the inside (i.e telcos, large public networks) - yes there is "bandwidth usage as the marginal opportunity cost" argument, but that is not the constraining factor. 

The constraining factor is their cost/value/RoI models, how the view the products they market and the (almost complete) inability of senior decision makers to be able to engage with the emergent properties of packet statistical multiplexing….

Its not about the money, its about the aspirations and expectations.


On 19 Sep 2012, at 09:20, Jon Crowcroft <Jon.Crowcroft at cl.cam.ac.uk> wrote:

> So here's an idea - (pace, Bob Briscoe and Google Adwords)
> Decongestant Adverts 
> (DA - Like
> Congestion Exposure, only backwards - employing Yet Another Level of Redirection called
> Re-Re-ECN...
> bandwidth doesn't have much operational cost - te real cosrt is the shadow price of other people's traffic you displace - if there isn't other traffic, then the additional cost of carrying yours is little.
> So we can have a receiver pays model for capacity - and the way they pay is via third party ads..
> now this works very nicely if we observe that
> congesiton exposire requires you to transparently reveal where the congestion is - i.e. the source of ECN marks...
> so the source can also reflect  the receiver to a wiling advertiser site, who then sends adverts with ECN-willing-to-pay marks ...
> sine the adverts flow the opposite direction from the traffic they don't add to congestion - indeed on many links (e.g. Adsl) there's plenty of capacity that way anyway
> that way, the net is free at the network layer, not just uo in the clouds
> what say?
> I see a bright new decongested future, full of IP banner ads
> j.
> On Wed, Sep 19, 2012 at 12:25 AM, Fred Baker (fred) <fred at cisco.com> wrote:
> On Sep 18, 2012, at 3:26 PM, Arjuna Sathiaseelan wrote:
> > Lets put the economics aside for a moment. I am more thinking like if
> > we can assign a class of IP addresses, where essential government
> > services  run, and lets say if the intermediate network devices are
> > configured (within the network operators) to recognise that these IP
> > addresses can be allowed to access without the client/user to pay,
> > then the network operators can always allow access to these services.
> > So are there any technological challenges here to realize this? I dont
> > think so.
> There's no technical challenge there. It's a business problem. Allocate some addresses from the existing pool and use them for a defined service such as you're describing.
> What happens next, of course, is that since bandwidth costs money and no money is being exchanged, one gets no bandwidth. You've had the experience in hotels, no doubt; they offer free wifi in every room, by which they mean they have installed wifi APs on a LAN and connected that to some service provider. It works just fine as long as you send no packets on it. If you decide to send packets, oh, well gee. 20% loss is not a problem, is it? It's better than losing ALL of the packets, and after all it's free...

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.postel.org/pipermail/end2end-interest/attachments/20120919/91a09ef0/attachment-0001.html

More information about the end2end-interest mailing list