Re: [multipathtcp] MPTCP carrying UDP

Olivier Bonaventure <> Thu, 24 November 2016 11:25 UTC

Return-Path: <>
Received: from localhost (localhost []) by (Postfix) with ESMTP id 2BA651293F5 for <>; Thu, 24 Nov 2016 03:25:28 -0800 (PST)
X-Virus-Scanned: amavisd-new at
X-Spam-Flag: NO
X-Spam-Score: -4.321
X-Spam-Status: No, score=-4.321 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, RCVD_IN_DNSWL_MED=-2.3, RCVD_IN_MSPIKE_H3=-0.01, RCVD_IN_MSPIKE_WL=-0.01, SPF_PASS=-0.001] autolearn=ham autolearn_force=no
Authentication-Results: (amavisd-new); dkim=pass (1024-bit key)
Received: from ([]) by localhost ( []) (amavisd-new, port 10024) with ESMTP id SnG1k9_Hx0Xj for <>; Thu, 24 Nov 2016 03:25:26 -0800 (PST)
Received: from ( []) (using TLSv1.2 with cipher AECDH-AES256-SHA (256/256 bits)) (No client certificate requested) by (Postfix) with ESMTPS id AD06E129AEB for <>; Thu, 24 Nov 2016 03:25:03 -0800 (PST)
Received: from ( []) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) (Authenticated sender: by (Postfix) with ESMTPSA id 878C267DCBA; Thu, 24 Nov 2016 12:24:54 +0100 (CET)
DKIM-Filter: OpenDKIM Filter v2.9.2 878C267DCBA
DKIM-Signature: v=1; a=rsa-sha256; c=simple/simple;; s=selucl; t=1479986694; bh=2STtPkOBWO1iLGs80up5g/TqFVo9q8+MU/8LatlSM6g=; h=Reply-To:Subject:References:To:Cc:From:Date:In-Reply-To; b=pfyVgqtQzqri81caybDMJEkZThn+EXDnR0P5nNbrmCqEWpwkLgSyTkCRqSPsjWP1u CRDwD8U+Z0oNRFGboKRXwfnKpUFzgKjvvaFG8Du5Jht0uiSbdNv18ePY/6CC5vF8jZ jHntgscjB1P5pf9Tn3B+SX2D17aWbuuOor66fEwU=
X-Virus-Status: Clean
X-Virus-Scanned: clamav-milter 0.99.2 at smtp-2
References: <> <20161123120926.7ed52bd4@sne-UX31E> <787AE7BB302AE849A7480A190F8B933009DB7E96@OPEXCLILMA3.corporate.adroot.infra.ftgroup> <> <787AE7BB302AE849A7480A190F8B933009DB856B@OPEXCLILMA3.corporate.adroot.infra.ftgroup>
To:, =?UTF-8?Q?S=c3=a9bastien_Noel?= <>, "" <>
From: Olivier Bonaventure <>
Message-ID: <>
Date: Thu, 24 Nov 2016 12:24:53 +0100
User-Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.11; rv:45.0) Gecko/20100101 Thunderbird/45.5.0
MIME-Version: 1.0
In-Reply-To: <787AE7BB302AE849A7480A190F8B933009DB856B@OPEXCLILMA3.corporate.adroot.infra.ftgroup>
Content-Type: text/plain; charset=utf-8; format=flowed
Content-Transfer-Encoding: 7bit
X-Sgsi-Spamcheck: SASL authenticated,
X-SGSI-MailScanner-ID: 878C267DCBA.A649A
X-SGSI-MailScanner: Found to be clean
X-SGSI-Spam-Status: No
Archived-At: <>
Cc: "" <>
Subject: Re: [multipathtcp] MPTCP carrying UDP
X-Mailman-Version: 2.1.17
Precedence: list
List-Id: Multi-path extensions for TCP <>
List-Unsubscribe: <>, <>
List-Archive: <>
List-Post: <>
List-Help: <>
List-Subscribe: <>, <>
X-List-Received-Date: Thu, 24 Nov 2016 11:25:28 -0000


>>>> To understand the interactions between QUIC and an underlying MPTCP
>>>> transport, we performed some experiments by running QUIC over OpenVPN
>>>> that runs itself over an MPTCP connection. This is the closest scenario
>>>> to what you are discussing based on existing open-source software.
>>> [Med] The schema we are investigating does not include this cascaded
>> layers. It is only UDP payload transported in plain MPTCP connections.
>> The cascading layers increase the CPU and byte overhead, but the
>> interactions between the congestion control schemes and the reliablity
>> mechanisms remains.
> [Med] My point is that we need to isolate the impact of those cascaded layers on the CPU/Performance efficiency from pure interactions between the CCs.

Given the performance of the PCs used as router, the CPU has no impact here.

The lessons that we've learned from this experiment, and others where 
users have reported experiments with running TCP over OpenVPN running 
over MPTCP is that it is a bad idea to run a protocol that uses a 
congestion control scheme above a protocol that provides a reliable 
delivery before the reliable delivery comes from retransmissions and 
those retransmissions hide packet losses that serve as indication for 
the congestion controller. This lessons applies to QUIC and similar 

If you have other experimental data, please share it on the list so that 
we can discuss based on real data.