[multipathtcp] Load balancing MPTCP

Costin Raiciu <c.raiciu@cs.ucl.ac.uk> Tue, 31 May 2016 08:35 UTC

Return-Path: <c.raiciu@cs.ucl.ac.uk>
X-Original-To: multipathtcp@ietfa.amsl.com
Delivered-To: multipathtcp@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id D223612B04F for <multipathtcp@ietfa.amsl.com>; Tue, 31 May 2016 01:35:23 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -5.626
X-Spam-Level:
X-Spam-Status: No, score=-5.626 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, RCVD_IN_DNSWL_MED=-2.3, RP_MATCHES_RCVD=-1.426] autolearn=ham autolearn_force=no
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 fl2HYMFywzrr for <multipathtcp@ietfa.amsl.com>; Tue, 31 May 2016 01:35:21 -0700 (PDT)
Received: from bells2.cs.ucl.ac.uk (bells2.cs.ucl.ac.uk [128.16.5.33]) (using TLSv1 with cipher DHE-RSA-AES256-SHA (256/256 bits)) (No client certificate requested) by ietfa.amsl.com (Postfix) with ESMTPS id 89BDB12B04E for <multipathtcp@ietf.org>; Tue, 31 May 2016 01:35:21 -0700 (PDT)
Received: from [141.85.233.104] (helo=[172.19.9.152]) by bells2.cs.ucl.ac.uk with esmtpsa (TLSv1:AES256-SHA:256) (C.Raiciu authenticated) (Exim 4.54) id 1b7f8u-0005MQ-Pc; Tue, 31 May 2016 09:35:16 +0100
Content-Type: text/plain; charset="utf-8"
Mime-Version: 1.0 (Mac OS X Mail 8.2 \(2104\))
From: Costin Raiciu <c.raiciu@cs.ucl.ac.uk>
In-Reply-To: <4E6D9951-D1EB-4F1F-8666-7189357053C0@gmail.com>
Date: Tue, 31 May 2016 11:35:15 +0300
Content-Transfer-Encoding: quoted-printable
Message-Id: <AFFED261-F6F0-45AC-A54B-4110AE81C9C7@cs.ucl.ac.uk>
References: <4E6D9951-D1EB-4F1F-8666-7189357053C0@gmail.com>
To: Alan Ford <alan.ford@gmail.com>
X-Mailer: Apple Mail (2.2104)
Archived-At: <http://mailarchive.ietf.org/arch/msg/multipathtcp/r5rffL3G93a64Xj3Gb43hT2gYYE>
Cc: multipathtcp <multipathtcp@ietf.org>
Subject: [multipathtcp] Load balancing MPTCP
X-BeenThere: multipathtcp@ietf.org
X-Mailman-Version: 2.1.17
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: <https://mailarchive.ietf.org/arch/browse/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, 31 May 2016 08:35:24 -0000

Dear all,

We have published a paper that presents the design and implementation of a MPTCP load balancer for datacenters.  The paper is called Datacenter Scale Load Balancing for Multipath Transport, will appear in HotMiddlebox 2016 and you can download it here: http://nets.cs.pub.ro/~costin/files/mplb.pdf

In short, we use the unmodified MPTCP protocol but encode the server identifier in the least significant 10bits of the timestamps. This is a hack and we acknowledge it, but it shows that one can build scalable load balancers without changing the protocol.

We are now working on a refinement of our design that will not rely on timestamps. Following Olivier’s suggestion, we will use destination port numbers for subsequent subflows for load balancing purposes (as you know, destination port numbers don’t really matter for subflow setup where the token is used instead). We will write a draft about our solution and present it at the Berlin IETF.

What this work shows is that there is no need to (drastically or at all) change the MPTCP protocol for load balancer support.

Best.
Costin