Re: [p2pi] [p2prg] Mythbustering P2P traffic localization

Arnaud Legout <> Fri, 27 February 2009 10:44 UTC

Return-Path: <>
Received: from localhost (localhost []) by (Postfix) with ESMTP id 8AD4C3A6A67; Fri, 27 Feb 2009 02:44:08 -0800 (PST)
X-Virus-Scanned: amavisd-new at
X-Spam-Flag: NO
X-Spam-Score: -6.249
X-Spam-Status: No, score=-6.249 tagged_above=-999 required=5 tests=[BAYES_00=-2.599, HELO_EQ_FR=0.35, RCVD_IN_DNSWL_MED=-4]
Received: from ([]) by localhost ( []) (amavisd-new, port 10024) with ESMTP id LekC1SnZVcvX; Fri, 27 Feb 2009 02:44:07 -0800 (PST)
Received: from ( []) by (Postfix) with ESMTP id 4674C3A6803; Fri, 27 Feb 2009 02:44:06 -0800 (PST)
X-IronPort-AV: E=Sophos;i="4.38,275,1233529200"; d="scan'208";a="35837197"
Received: from (HELO []) ([]) by with ESMTP/TLS/DHE-RSA-AES256-SHA; 27 Feb 2009 11:44:26 +0100
Message-ID: <>
Date: Fri, 27 Feb 2009 11:44:23 +0100
From: Arnaud Legout <>
User-Agent: Thunderbird (Windows/20081209)
MIME-Version: 1.0
To: Enrico Marocco <>
References: <>
In-Reply-To: <>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 8bit
Cc: "" <>,, "" <>, "" <>
Subject: Re: [p2pi] [p2prg] Mythbustering P2P traffic localization
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: P2P Infrastructure Discussion <>
List-Unsubscribe: <>, <>
List-Archive: <>
List-Post: <>
List-Help: <>
List-Subscribe: <>, <>
X-List-Received-Date: Fri, 27 Feb 2009 10:44:08 -0000

[Should we limit the cross posting? I believe sending this thread to 4 
mailing lists is too much
(may be this is because I am subscribed to all four)]


here is some feeback based on what we did in [Le Blond].

section 3
We show that we can achieve much higher savings than 80% of reduction. 
The direct implication of this
result is that you can still achieve significant savings with few peers 
per ISP (per torrent).

section 4
we explain the relation between peers download completion time and 
network congestion.
With large enough network congestion you can significantly improve the 
peers download completion time.
Therefore, it is likely that the different gains shown by field 
experiments report different level of congestion
of the underlying network during the experiments.

section 8.1
you write that we did simulations. We did not, we did experiments with 
real BitTorrent clients.

section 8.2
it is not clear what you mean in "as shown in [Le Blond], the right 
balance of randomness and
  locality depends on the P2P algorithm". We did not explore several P2P 
protocols, but only BitTorrent.
We show that BT is extremely robust to high locality (as long as there 
is enough randomness within each ISP).
We cannot conclude for any other P2P protocol.

"On the other hand, P2P systems not adopting the tit-for-tat approach
  (e.g. the eDonkey network) should not be damaged by locality-based"
It is not clear to me. Peer selection (tit-for-tat like choking algo in 
BT) is not the only one factor of efficiency, there is also piece 
Edonkey/emule also use rarest first (but a less efficient peer selection 
algorithm). Therefore, as locality adversely impact
piece diversity, my bet is that Edonkey/emule will suffer from locality.


Enrico Marocco wrote:
> Hello folks,
> we have just submitted a draft that tries to summarize many discussions
> about possible effects (and side-effects) of P2P traffic localization:
> The document is very early and the conclusions may be controversial; any
> comments, feedback and contributions to improve it will be greatly
> appreciated.
> Apologies for cross-posting, I'm sending this email to all the lists
> where some of the discussions happened; please consider addressing any
> follow-up to p2prg only.
> ------------------------------------------------------------------------
> _______________________________________________
> p2prg mailing list

Arnaud Legout, Ph.D.

INRIA Sophia Antipolis - Planète  Phone :
2004 route des lucioles - BP 93   Fax   :
06902 Sophia Antipolis CEDEX      E-mail:
FRANCE                            Web   :