[LINK] RFI: P2P Traffic 2005-06

[rchirgwin at ozemail.com.au]

Apologies for thread-skipping - I read the other posts to this, and they 
triggered a couple of lines of thought.

The first is, as Roger, Kim et al have observed, definition is 
everything. The trick of the researchers here is to say "here's the 
traffic, here's what I define as P2P, therefore P2P = X%". Market 
research is continually guilty of this - if the data isn't suitable, 
change the definitions (eg: X-Box 360 failed to dominate console sales, 
so the market analysts redefined the market. Now there's a new category, 
"next generation consoles", which is dominated by 360).

The second is that most researchers are trying to classify Internet 
traffic according to end-user application. That may be valid (although 
in some ways I think not!), but this hints that the definition of P2P is 
not by the characteristics of host-to-host communications, but by the 
application generating those communications. Therefore you can change 
the amount of P2P traffic merely by changing the applications you include...

The third thought is this: once the user traffic has passed a given 
boundary (let's call it the point of ingress), isn't all traffic "peer 
to peer"? The definition being sought by marketers isn't a definition of 
"Internet traffic" but "access network traffic"...

Random musings which may be shredded at will...

RC

Roger Clarke wrote:

> I'm finalising this for publication:
>
> <warning: academic paper>
>     Towards a P2P Research Agenda
>     http://www.anu.edu.au/people/Roger.Clarke/EC/P2PRes.html
> </warning>
>
> In the process, I'm checking on the state of P2P traffic.
>
> This is about the best I've found:
> http://www.cachelogic.com/research/p2p2005.php
>
> I'm unclear how to get any data from http://www.bigchampagne.com/
>
> Are linkers aware of other sources that I should consult?
>
> Thanks!!
>