[LINK] Australian ISP Peering

[Geoff Huston]

On 17/03/2009, at 6:39 PM, stephen at melbpc.org.au wrote:

> Geoff Huston writes,
>
>> enough of this old fart stuff - back to work!  ;-) Geoff
>
> Great to see you Link, Geoff. You are, surely, an Aussie institution.
>

ooops! :-)


> Whilst Linkers will all note, and agree, with your IPv6 concerns, may
> one ask your opinions regards another issue which arose on Link today?
>
> That issue is, Australian ISP peering.
>
> At http://www.isp-planet.com/business/peering_australia.html you are
> quoted at some length, in the following Sep 5th 2000 item, re peering.
>

sigh - the problem with today's network is that nothing is ever  
allowed to fade away into history is it!

oh well.

Here's an updated view about peering....

Peering is certainly an art. True peering in this industry is stable  
as long as either party to a peering arrangement can walk away from  
the peering setup and the other party is not desperate to re-establish  
the peering relationship. i.e. peering is a relationship between  
equals in a market, rather than an effort to leverage the position of  
one by the other, and the true test of peering is that a true peering  
relationship is one that is based on a parity of lack of need for the  
peering relationship.

One could argue that some of the desperation in the flareups in the  
peering debates in this country over the years were illustrative that  
some of the players were either particularly ill-informed about the  
underlying dynamics of peering, or that it was all just smoke and  
mirrors and the underlying issue was, and still remains price. After  
all if the debate was truly between equals in the market then peering  
becomes a statement of mutual levels of self-interest being expressed,  
and there is no requirement to force one party to the table at the  
behest of the other. So, as I see it, the squeals of "unfair!" about  
one party not peering with another is in fact a clear and unambiguous  
public signal that peering was NOT the appropriate relationship for  
these two parties in the first case!

So what is behind this never-ending peering debate?

I suspect that its nothing to do with "peering" and more to do with  
some basic properties of this industry.

There is no doubt that this industry was, and still is, an industry  
that derives its greatest economy though scale and volume. Hundreds of  
small undercapitalized local businesses does not make an efficient  
national infrastructure, nor could such a landscape create solid  
foundations for a competitive economy. This business requires large  
scale capital investment and incredibly high volumes in order to  
leverage economies of scale. So what we have in Australia is no  
different to many other economies today: A small number of primary  
providers who have made large scale capital (and process) investments  
who provide an undifferentiated product on a large scale, and a  
peripheral constellation of resellers and customization agents who  
repackage and resell the product into various forms of niche market  
segments. In terms of a total offering these players need each other.  
The larger player is incapable of performing the high touch  
customization of the product and the smaller players lack the capital  
backing to make the necessary investments to enter the mass market to  
sustain a cost effective base product. But this mutual need is NOT a  
peering relationship, no matter how it gets dressed up. Its a  
relationship that is more akin to a value add relationship where the  
resellers need access to the base product, and the base producer needs  
the resellers to broaden the channels to market. As is always the case  
the form of relationship is unstable, however. Both sides see  
themselves being disadvantaged, and want to take a higher share of the  
revenue at the expense of the other. The reseller market in the  
Internet space appears to be quite vibrant and enjoys some levels of  
strong competition, but this can backfire on the reseller sector as  
the resellers often appear to compete more strongly with each other,  
and tend to lack a basic ability to band up to create a consolidated  
position to confront the base product providers to force the wholesale  
price down through collective action. So the base providers tend to  
use an approach which leverages the divisions and competition in the  
reseller market to their advantage. The same cannot be said of the  
base infrastructure market, and here we tend to teeter between glut  
and famine and rapid bursts of insane competition that undercut any  
longer term investment value in infrastructure and waves of  
monopolistic strangulation of the reseller market and price gouging of  
the end consumer.

So peering is not a panacea here - it has its uses and meets some  
objectives in this market, but it won't solve everything. This is a   
business where economies of scale are critical to sustaining a cost  
efficient infrastructure base, and if you attempt to perform this  
infrastructure provisioning with private sector vehicles you will  
always find that the base business will tend to aggregate into a small  
number of high volume operators. Now you could always attempt to  
perform the infrastructure investment using public funds (such as  
happened with the initial overland and undersea telegraph services in  
the late 19th century) but right now we are living out the legacy of  
the triumph of deregulation and competitive capitalism of the 80's so  
we have the challenge of attempting to fit a whole bunch of mutually  
incompatible concepts into a public policy framework about  
telecommunications service provision for the next few decades. Little  
wonder that it gets confusing from time to time, and little wonder  
that ISP players, both large and small, slide off into dark and  
unpalatable areas of market abuse from time to time.

I'm not sure if this diatribe has shed more light or confusion on the  
topic, but, you did ask.... :-)

regards,

    Geoff