[Air-l] peering post
Bram Dov Abramson
bda at bazu.org
Sun Mar 11 19:37:44 PDT 2007
Nancy Paterson <Nancy.Paterson at senecac.on.ca>, 11.03.07:
> Bram, you missed ENRON (the largest failure in corp history) its not
> clear to what extent their engaging in bandwidth trading contributed
> to their demise but it really is not relevant - ENRON was a
> significant corp failure. They created the first bandwidth trade.
No, I don't think it's accurate to say that Enron participated in "a
commodity market in packet-based bandwidth". From what I recall Enron
traded only dark and lit fibre (specifically, SONET). There was some
talk of Internet bandwidth, but I don't believe Enron ever went
forward with that. If you widen your ambit beyond "packet-based
bandwidth", by which I assume you intend "Internet bandwidth" -- ATM
and frame relay networks are packet-based, but not really relevant
here -- then you could take in the sorts of things Enron was doing.
Because Enron hoped that growth in demand for Internet bandwidth would
help create an environment in which the sort of underlying bandwidth
it was producing, selling, and buying could be commoditised, I think
this would remain true to your topic -- as long as the two network
layers in question were not conflated. They are different in a number
of important ways, particularly geographically: "raw" bandwidth
involves city pairs (point to point); Internet transit is usually
one-ended.
> transit IS peering - just for fee
That seems an unusual use of these terms to me -- generally they are
counterposed such that "peering" is a settlement-free form of transit
where, as the trade-off, traffic destined only to the peer is
delivered. In other words: pay me for me to deliver all traffic to
the rest of the Internet, and vice versa (transit); free-peer with me
to deliver all the traffic you've got that goes to my end-users, and
vice versa (peering).
Some providers make "paid peering" available, a sort of halfway step
-- ie: pay me less than transit to deliver all traffic to my
end-users, etc. See, for instance, AOL, who apparently does this
(which, given the nature of paid-peering and of AOL, respectively,
makes sense): <http://www.atdn.net/paid_peering.shtml>.
>> Don't get me wrong, though! This is one of a number of important areas to
>> look at. I question whether it is the most significant or even an
>> extremely significant cost driver, relative to others, in the
>> international context in which you want to operate -- I suspect that other
>> cost drivers, most of them related to state-sponsored chokeholds, tend to
>> be both more determinative, and highly connected, to this issue. But
>> there is lots to be said and lots being said on this topic, with a number
>> of interesting approaches in recent years to try and bring transparency to
>> this domain.
>
> Thanks for the book references
> I'm not looking at anti-trust but at transparency in peering agreements
> - and how this may be able to happen
Yes. I don't know that it is useful to cloister these from one
another quite so absolutely, though. Typically transparency in
bilateral commercial relations is either entered into at the behest of
the parties themselves, or mandated by some third party with varying
levels of suasion -- "in the shadow of", be it the shadow of a
community of network operators, best practice guidelines promulgated
by some formal association, shared memberships in a network facility
such as a multiparty Internet exchange or, of course, state
requirements.
Of the latter category, early and very significant third-party
attempts to provide trasparency came in the very situational context
of antitrust actions, particularly in the MCI WorldCom and Sprint
transactions. They helped set the stage for subsequent attempts,
particularly that which has moved forward in Australia. That's why
I'd think one would to include that history of antitrust enforcement
as part of the story. I have some references to these in a chapter of
Greg Elmer's _Critical Perspectives on the Internet_
(http://shorl.com/frifrebrudrebrine) if it is of help.
cheers
Bram
More information about the Air-L
mailing list