AOL & Cogent

Mike Leber mleber at
Mon Dec 30 07:07:46 UTC 2002

On Sun, 29 Dec 2002, Paul Vixie wrote:

> > The perceived "money on the table" frequently doesn't exist and attempts
> > to get it may produce the opposite result.
> well, yeah, sure, but...
> > * Who they shift the traffic to may be your competitor.
> least you know they are paying SOMEBODY, thus supporting the market
> you want to be in.  you can then compete in that market.  if everybody who
> could peer in N places worldwide could just get peering, then all kinds of
> per-bit revenue for "high tier" network owners would turn into per-port
> revenue for exchange point operators.  where's the market in that?  how
> could a "high tier" even exist in those conditions?

This is a straw man argument.  I could just point out how it's technically
wrong except that would be no fun so instead I'll give analogies first:

Your argument is like saying that everybody that has taken a wood shop
class in high school (or junior high) can build their own house, so home
builders are going to be out of business.

Or how about... Since everybody that has a truck can drive a package from
San Jose to New York, Federal Express and United Parcel Service are no
longer needed.

Or most mundane yet...  Everybody knows how to make sandwitches so there
aren't going to be any kind large sandwitch chains, let alone
multinational corporations that serve food.


Networks cost money to build and operate.  Geography, both physical and
political, provide for varying costs over different routes.  The majority
of large networks don't have the exact same routers in the exact same
places connected with the exact same circuits.  Operational costs, capital
costs, customer service attitudes, and policies are different between
companies.  All of these features define the specific value added of a
network.  Economic pressure and the underlying technology determine how
many companies can exist.

Even if there were low barriers to entry, it doesn't mean that there won't
be cases where it makes perfectly reasonable sense to some networks to
outsource part of their infrastructure needs.  This might be as simple as
coverage for a specific market or backup capacity.

> we will need to 1000X the traffic volume again before this stops
> working again. 
> which should take about a year.

Heh.  That should be interesting.  :)

In the long run capacity is likely to be capable of expanding at the rate
of moores law.

"Desired" bandwidth per end user appears to increase in a nonlinear manner
at the introduction of new protocols (i.e. HTTP, Kazaa).

If we have enough of these nonlinear transitions we might even someday be
able to make a moores law equivalent for end user bandwidth demand (the
chip industry has a few more years on us to be able to make empirical
conclusions regarding industry constants).

Then you could compare the curve of the end user bandwidth demand law to
the moores law curve and make interesting prognostications.

To illustrate how moores law and the hypothetical end user bandwidth
demand law are different, for anybody that has upgraded their personal
workstation to twice the processor speed or greater, to do the exact same
end user task (i.e. visit a website) the day after you upgraded did you
generate twice as much bandwidth?  probably not.


+----------------- H U R R I C A N E - E L E C T R I C -----------------+
| Mike Leber           Direct Internet Connections   Voice 510 580 4100 |
| Hurricane Electric     Web Hosting  Colocation       Fax 510 580 4151 |
| mleber at                              |

More information about the NANOG mailing list