Re: [multipathtcp] Collection of MPTCP proxy scenarios.

Lukasz Budzisz <lukasz.budzisz@tu-berlin.de> Tue, 05 May 2015 11:18 UTC

Return-Path: <lukasz.budzisz@tu-berlin.de>
X-Original-To: multipathtcp@ietfa.amsl.com
Delivered-To: multipathtcp@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 447341A8903 for <multipathtcp@ietfa.amsl.com>; Tue, 5 May 2015 04:18:21 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -3.859
X-Spam-Level:
X-Spam-Status: No, score=-3.859 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, HELO_EQ_DE=0.35, HTML_MESSAGE=0.001, RCVD_IN_DNSWL_MED=-2.3, T_RP_MATCHES_RCVD=-0.01] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id Ur2ZXrrCEtEu for <multipathtcp@ietfa.amsl.com>; Tue, 5 May 2015 04:18:16 -0700 (PDT)
Received: from mail.tu-berlin.de (mail.tu-berlin.de [130.149.7.33]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by ietfa.amsl.com (Postfix) with ESMTPS id 002BB1A88F6 for <multipathtcp@ietf.org>; Tue, 5 May 2015 04:18:15 -0700 (PDT)
X-tubIT-Incoming-IP: 91.65.35.1
Received: from ip5b412301.dynamic.kabel-deutschland.de ([91.65.35.1] helo=[192.168.1.128]) by mail.tu-berlin.de (exim-4.76/mailfrontend-6) with esmtpsa [UNKNOWN:AES128-SHA:128] for <multipathtcp@ietf.org> id 1Ypard-00084x-45; Tue, 05 May 2015 13:18:14 +0200
Message-ID: <5548A6F4.5080808@tu-berlin.de>
Date: Tue, 05 May 2015 13:18:12 +0200
From: Lukasz Budzisz <lukasz.budzisz@tu-berlin.de>
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:31.0) Gecko/20100101 Thunderbird/31.6.0
MIME-Version: 1.0
To: multipathtcp@ietf.org
References: <C5C3BB522B1DDF478AA09545169155B46E336D13@nkgeml507-mbx.china.huawei.com> <553DF29E.7000502@tu-berlin.de> <C5C3BB522B1DDF478AA09545169155B46E337225@nkgeml507-mbx.china.huawei.com> <80C0017654A043479F53C41112BE847687D93BC532@WP40046.corp.ads> <C5C3BB522B1DDF478AA09545169155B46E337A2F@nkgeml507-mbx.china.huawei.com>
In-Reply-To: <C5C3BB522B1DDF478AA09545169155B46E337A2F@nkgeml507-mbx.china.huawei.com>
Content-Type: multipart/alternative; boundary="------------050802000004070803030100"
X-PMX-Version: 6.0.0.2142326, Antispam-Engine: 2.7.2.2107409, Antispam-Data: 2015.5.5.110017
X-PMX-Spam: Gauge=IIIIIII, Probability=0%, Report=''
Archived-At: <http://mailarchive.ietf.org/arch/msg/multipathtcp/Mg7LY_agfpTLkMMTxOp1apGZvAw>
Subject: Re: [multipathtcp] Collection of MPTCP proxy scenarios.
X-BeenThere: multipathtcp@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Multi-path extensions for TCP <multipathtcp.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/multipathtcp>, <mailto:multipathtcp-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/multipathtcp/>
List-Post: <mailto:multipathtcp@ietf.org>
List-Help: <mailto:multipathtcp-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/multipathtcp>, <mailto:multipathtcp-request@ietf.org?subject=subscribe>
X-List-Received-Date: Tue, 05 May 2015 11:18:21 -0000

Hi Xingpeng,

sorry for a delayed reply due the holidays.
Our draft proposes a proxy architecture that is agnostic to the 
multipath solution, meaning it may be applied for multipath TCP (as a 
multipath-enabling solution) as well.
More importantly however, the use case scenarios identified in that 
draft, apply for multipath TCP proxies, too.

Best regards,
Lukasz

On 05/05/15 03:50, Weixinpeng (Jackie) wrote:
>
> Hi Jordan,
>
> Sorry for late reply. (I have just come back from a vacation..)
>
> I think you have provided a very good use case related with the home 
> CPE devices that have more than one interfaces.
>
> As my understanding, this use case focuses on providing multipath 
> support in the network for TCP hosts. My feeling is
>
> that this use case is a bit of similar to what Lukasz provided below. 
> I think this kind of back-to-back MPTCP proxy setup might
>
> be useful for specific scenario.
>
> Regards,
>
> Xinpeng
>
> *From:*multipathtcp [mailto:multipathtcp-bounces@ietf.org] *On Behalf 
> Of *Jordan Melzer
> *Sent:* Friday, May 01, 2015 1:13 AM
> *To:* multipathtcp@ietf.org
> *Subject:* Re: [multipathtcp] Collection of MPTCP proxy scenarios.
>
> Hi Xinpeng,
>
> I have used a back-to-back MPTCP proxy setup to allow TCP to multipath 
> over a segment of network.  My use case is allowing a home router to 
> use more than one Internet connection.  The home router would have one 
> MPTCP proxy and the ISP would have another.  The home devices would 
> not need to run MPTCP, nor would any Internet servers. The advantage 
> of MPTCP proxies here is that the different network connections coming 
> into the home need not have deterministic capacities, and the system 
> supports smooth failover as well as aggregation.  DSL, LTE, cable, and 
> WiFi (eg, from a neighbouring home) are all candidate technologies to 
> be aggregated.
>
> The challenge in this configuration is for the proxy to not prevent a 
> native MPTCP device from running an end-to-end MPTCP connection.  The 
> default proxy one gets by running a TCP proxy, eg HAPROXY, on a 
> machine running an MPTCP kernel doesn’t avoid proxying MP_JOINs from 
> sessions it doesn’t know about and doesn’t keep the keys the same on 
> both sides of the connection, preventing an MPTCP capable home client 
> from successfully doing an MP_JOIN using a path that doesn’t go 
> through the home proxy.
>
> One could probably modify HAPROXY to do this behaviour or make 
> nftables rules that make MPTCP traffic bypass the proxies altogether.
>
> I am not entirely satisfied that this kind of setup is a great idea, 
> but the use case of bonding different connections together without 
> changing anything at the ends of the network could be compelling.
>
> Jordan
>
> *From:*multipathtcp [mailto:multipathtcp-bounces@ietf.org] *On Behalf 
> Of *Weixinpeng (Jackie)
> *Sent:* April 28, 2015 11:14 PM
> *To:* lukasz.budzisz@tu-berlin.de 
> <mailto:lukasz.budzisz@tu-berlin.de>; multipathtcp@ietf.org 
> <mailto:multipathtcp@ietf.org>
> *Subject:* Re: [multipathtcp] Collection of MPTCP proxy scenarios.
>
> Hi Lukasz,
>
> Thanks for your information.
>
> I have read through the document, and find it an interesting solution, 
> but I think the solution in the document is
>
> about how to transmit traditional TCP traffic through multipath, and 
> it is a bit of different from MPTCP protocol.
>
> Thanks.
>
> Regards,
>
> Xinpeng
>
> *From:*Lukasz Budzisz [mailto:lukasz.budzisz@tu-berlin.de]
> *Sent:* Monday, April 27, 2015 4:26 PM
> *To:* Weixinpeng (Jackie)
> *Subject:* Re: [multipathtcp] Collection of MPTCP proxy scenarios.
>
> Dear Xinpeng,
> you may have a look at the draft we presented in the IETF Paris meeting:
> https://tools.ietf.org/html/draft-ayar-transparent-sca-proxy-00
> Best regards,
> Lukasz
>
> ------------------------------------------
> Łukasz Budzisz, Ph.D.
> Research Fellow
> Technische Universität Berlin (TU Berlin)
> Telecommunication Networks Group
> Tel: +49 30 314 23836
> ------------------------------------------
>
> On 27/04/15 10:20, Weixinpeng (Jackie) wrote:
>
>     Hi all,
>
>     MPTCP proxy aims to provide additional deployment support for
>     MPTCP protocol, and the functions provided
>
>     by MPTCP proxy could involve several different aspects such as
>     supporting establishment of MPTCP connection between
>
>     MPTCP host and traditional TCP host, aggregating of subflows to
>     one point for security reasons,…
>
>     I am trying to figure out as many MPTCP proxy deployment scenarios
>     as possible to have a more deep understanding
>
>     of how MPTCP proxy should work. As an example, here are some
>     potential scenarios:
>
>     (1) MPTCP proxy in operator’s network. The operator could deploy
>     MPTCP proxy to assist MPTCP-capable UE to communicate
>
>     with TCP server on Internet using MPTCP.
>
>     (2) MPTCP proxy in enterprise network. Enterprise network could
>     ask all the traffic pass through security check point for security
>     reasons,
>
>     in this case, all the subflows belong to a MPTCP connection should
>     be aggregated at the same point.
>
>     Besides, some existing documents such as draft-deng-mptcp-proxy-01
>     and draft-lopez-mptcp-middlebox-00 also provide some related
>     discussion.
>
>     So if you know some additional MPTCP proxy deployment scenarios
>     especially the ones that have some special requirements, It will
>     be appreciated if you could
>
>     share it here. Or if you have any other considerations on the
>     MPTCP proxy topic, we can also discuss it here.
>
>     Thanks!
>
>     Regards,
>
>     -Xinpeng
>
>     _______________________________________________
>
>     multipathtcp mailing list
>
>     multipathtcp@ietf.org  <mailto:multipathtcp@ietf.org>
>
>     https://www.ietf.org/mailman/listinfo/multipathtcp
>
>
>
> _______________________________________________
> multipathtcp mailing list
> multipathtcp@ietf.org
> https://www.ietf.org/mailman/listinfo/multipathtcp