Hi Arnaud,
I think we both made our points clearly and I think we can both agree
that we could continue this discussion infinitely.
Since this group is not the right place to discuss particular
scientific papers in details, I suggest we stop right now.
Regards,
Maciek Wojciechowski
On Dec 4, 2008, at 6:44 PM, Arnaud Legout wrote:
Hi Maciej,
Maciej Wojciechowski wrote:
You are right, that a control experiment like yours can be very
helpful in getting a better understanding of what is in fact going
on.
I'm talking about Marshall's sentence from first post in this
thread "This would be a very encouraging reduction of inter-ISP
traffic if it could be approached in practice by what we are doing
here". I think that if those results could be reproduced it would
be great. I'm just highly doubtful that they can.
We do not intend to overstate our results. We are simply performing
a study intended to
give a first answer to the three questions we present in the
introduction:
-How far can we push locality?
-What is, at the scale of a torrent, the reduction of traffic that
can be achieved with locality?
-Can locality significantly deteriorate the peers experience?
I don't know what you mean by reproducibility. If you mean that we
will not find in the Internet
a torrent with the exact same characteristics than the ones we have
evaluated, you are right.
However, our results show that we can push locality further than
what was done up to now
with a great reduction on the inter-ISP traffic. Claiming that this
reduction will be 2 orders of magnitude
on a real torrent, or 1.75 or whatever you want would be pointless.
It will be much better
than what you can obtain with the locality values considered up to
now (around 80%), without a
negative impact on peers download completion time. This is the
important
point.
like? Why is it 20kB/s and not 200kB/s or 1037kB/s? Why 100MB and
not 257MB or 2057MB? I have never heard of inter ISP link with
2000kB/s capacity (not mentioning the 40kB/s). I have heard about
10Gbit/s ones though. I cannot see them anywhere in your work...
Few things that in my opinion are significant and are not taken
into consideration:
- other web activities (like web browsing, gaming, ftp,
youtube,...) of the users
- setting arbitrary max download and max upload speeds
- downloading many torrents at the same time
- vast part of real-world users are hidden behind a NAT
- users with very big symmetric links (e.g. 100Mbit) that are
seeding much more than downloading. Yes, there are such users...
- and many, many more.
Honestly, there are hundreds of factors, many of them very complex.
Do you really want to argue that the simplified model you presented
is good enough to reason about bittorent behavior?
This is exactly the point. There are so many factors that you cannot
test everything.
Either you measure real torrents and traffic, but it is only
relevant for what you have measured at that moment,
and you will never understand this way the impact of some specific
parameters. We did that for BT in another
context (see our IMC'2006 study)
and know that this is just one part of the big picture that allows
you to understand what is going on.
Or, you make controlled experiments varying one parameter at a time.
The question here is how you chose the
fixed parameters and the range of what you vary. This is based on
experience and understanding you got from
in the wild experiments, and related work.
We are quite confident that our choices are relevant and allow us to
show what we wanted to show.
But of course it is always possible to make different choices and to
compare different approaches.
This is why research is great.
I want one thing to be clear: I really like your paper, as a study
work that can help us understand the problem better. But with all
due respect, I simply cannot agree that this is a valid bittorrent
model. Hence, I don't see why those results would give us knowledge
about how the deployed ALTO system would perform in the real-world.
All depends on what you mean by knowledge. Don't attribute to our
study a larger scope than what we described.
Our study simply says that if you increase the locality up to very
high values
then the inter-ISP traffic will be lower without negative impact on
end-users.
Forget about absolute values, they are only there to compare
different experiments in the context of our study.
If you claim that we haven't shown that it makes sense to consider
high locality values, then I have to disagree.
Regards,
Arnaud.
_______________________________________________
alto mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/alto
