[ppml] Free Market

[William Herrin]

On 8/27/07, Iljitsch van Beijnum <iljitsch at muada.com> wrote:
> Although this changes some things I don't think it's
> a fundamentally different economic model. You still have peering and
> transit the way we have today.

Iljitsch,

Here's the economic problem you haven't adequately addressed:

NL gets a regional route. Lets say 10.128.0.0/10
US gets a regional route. Lets say 10.192.0.0/10

Fred and George are NL customers with 10.128.0.0/24 and 10.128.1.0/24
respectively.
Sam and Tom are US customers with 10.192.0.0/24 and 10.192.1.0/24 respectively.

Alice and Becky are NL ISPs connecting Fred and George respectively.
Christie and Debbie are US ISPs connecting Sam and Tom respectively.

Jenny and Kim are transatlantic providers. Jenny connects Alice and
Christie to each other while Kim connects Becky and Debbie to each
other.

Alice and Becky peer. Alice announces 10.128.0.0/24 to Becky. Becky
announces 10.128.1.0/24 to Alice.
Alice announces 10.128.0.0/10 to Jenny. Becky announces 10.128.0.0/10 to Kim.
Christie and Debbie peer. Christie announces 10.192.0.0/24 to Debbie
and Debbie announces 10.192.1.0/24 to Christie.
Christie announces 10.192.0.0/10 to Jenny. Debbie announces
10.192.0.0/10 to Kim.

This is a complete routing structure that works on a technical level
as envisioned in your geoaggregation paper.

Fred at 10.128.0.1 sends a packet to Tom at 10.192.1.1. Fred hands the
packet to Alice. Alice hands the packet to Jenny. Jenny hands the
packet to Christie. Christie hands the packet to Debbie. Debbie hands
the packet to Tom.

Fred pays Alice to transmit his packets.
Alice pays Jenny to transmit Fred's packets.
Tom pays Debbie to receive his packets.

WHO PAYS CHRISTIE to transmit or receive packets associated with Tom and Fred?


In every peering relationship on the Internet today, someone pays.
Traffic doesn't come in one peer and then out another. Traffic never
ever comes in a paid link and then out a reciprocal peer. For any
given packet, at least one of the endpoints is associated with either
the entity itself or one of its paying customers.

Your proposal breaks this economic model and offers nothing with which
to replace it.


If that isn't problem enough, what happens if Christie and Jenny are
low cost bottom feeders with bad packet loss while Debbie and Kim are
a high-cost, high quality providers?  Even though Tom is paying for
quality, Fred won't pass his packets to George and then Kim... Unless
they all ignore the aggregate routes and announce the longer prefixes
too, making the whole thing a colossal waste of effort.

Regards,
Bill Herrin

-- 
William D. Herrin                  herrin at dirtside.com  bill at herrin.us
3005 Crane Dr.                        Web: <http://bill.herrin.us/>
Falls Church, VA 22042-3004