the Union makes us Strong

Mon Oct 28 01:46:25 UTC 1996

William Allen Simpson" <wsimpson at greendragon.com> wrote:

>Several have noted in the past that Vadim has a religious predeliction
>toward "robber baron" oligopolies.  We've had plenty of historic
>experience with them here in the US.

Hm.  You never tried to live in a place without antitrust laws.
I bet you'd have quite different perspective then.

>The wolves have learned to eat grass (IP) in addition to their old fare
>(overpriced direct links, X.25, and OSI).

No, they don't.  Not while the bellheads are telco's kings of the
dumps.  I would venture to guess that telcos could have Internet for
lunch if they were willing to sacrifice some corporate sacred cows.
I learned that hard way -- remember which telco managed to get viable
Internet business first?   It wasn't that hard to make some competitors
existing at the time very unhappy -- all it took is one engineer who
can deal with socialism and one manager who was willing to bend the
rules.

I could also tell why there's no significant competition from telcos in
the dial-up field, but i'd rather prefer not to make more personal
enemies.

>Gentlefolk, if we are to avoid the doom that Vadim predicts, the
>large number of rabbits need to band together to fend off the wolves.

No.  Then "rabbits" will become about as good as the wolves.  The role
of small business is to move fast and skim the cream.  The role of
big business is to exploit economies of scale.  Internet is no longer
an esoteric technology; so i don't see a reason why it cannot be done
by telcos.

>There was a great deal of talk among small providers in the halls and
>restaurants around NANOG last week.  Several different models of
>cooperation were discussed.

Right.  How about actually putting fiber in the ground?  Unless small
ISPs have their own data transmission facilities they have zero chance
to win in the long run.

>As I raised as an issue with the large providers during their
>presentation, the very nature of IP statistical multiplexing improves
>throughput when several providers join together and use the same pipe.

Done some reading on basics of queuing theory, didn't you?

Unfortunately, the theory is dry and tree of life is green -- i hate to
point out obvious but stochastic multiplexing does not work for self-similar
traffic like it does for Poission.  Things are a lot more interesing when
you've got traffic with infinite interarrival time variance.  No matter
how much you aggregate you still get the same stuff.

>Yet, if we are to survive, we will need to build a shared cooperative
>infrastructure tailored to our own needs.  The writing is on the wall.

Right on spot.  Can I suggest some reading of Lenin?  He dwelt quite a
lot on cooperation and collectivism.  That didn't work in reality, though.

>I propose that ISPs join together, build, and fund regional and
>metropolitan exchanges, and each of these exchanges make a group
>purchase of transit to other exchanges, instead of each ISP purchasing
>their own transit.  Eventually, the independent exchanges together
>will constitute a super provider which meets the draconian peering
>requirements promulgated by the big providers to limit competition.

You either end up with franchises or a large corporation taking over.
Been there, done that.  You may want to look at history of Relcom, which
was exactly what you descibe there.  Didn't last long that way, though.

>I believe that we need an umbrella organization to make this happen.
>There is apprehension that the CIX might involve conflict of interest
>among its membership.  The new ISPC might be a good choice.

CIX would be better off not providing any service in competition to its members.
If an entity is forwarding packets it _is_ in competition with its
"customers" or members, unless it is in it's chapter to disallow membership
of anybody who pays fees.   When you do that it's no different from a
cooperative, an economical form known to be unstable and prone to "spontaneous"
decay into a conventional corporation.

--vadim