From mboxrd@z Thu Jan 1 00:00:00 1970 From: Andy Gospodarek Subject: Re: [patch net-next V8] net: introduce ethernet teaming device Date: Mon, 14 Nov 2011 20:56:17 -0500 Message-ID: <20111115015616.GA25132@gospo.rdu.redhat.com> References: <1321085808-6871-1-git-send-email-jpirko@redhat.com> <4EC160B4.2040709@hp.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: Jiri Pirko , netdev@vger.kernel.org, davem@davemloft.net, eric.dumazet@gmail.com, bhutchings@solarflare.com, shemminger@vyatta.com, fubar@us.ibm.com, andy@greyhouse.net, tgraf@infradead.org, ebiederm@xmission.com, mirqus@gmail.com, kaber@trash.net, greearb@candelatech.com, jesse@nicira.com, fbl@redhat.com, benjamin.poirier@gmail.com, jzupka@redhat.com, ivecera@redhat.com To: Rick Jones Return-path: Received: from mx1.redhat.com ([209.132.183.28]:2845 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752544Ab1KOB4k (ORCPT ); Mon, 14 Nov 2011 20:56:40 -0500 Content-Disposition: inline In-Reply-To: <4EC160B4.2040709@hp.com> Sender: netdev-owner@vger.kernel.org List-ID: On Mon, Nov 14, 2011 at 10:40:52AM -0800, Rick Jones wrote: > On 11/12/2011 12:16 AM, Jiri Pirko wrote: >> This patch introduces new network device called team. It supposes to be >> very fast, simple, userspace-driven alternative to existing bonding >> driver. > > What is the definition of "very" here - relative to bonding I presume? > Are there actual performance figures available at this point? Something > along the lines of test through both on the same hardware. I don't have > HW on which to run myself but would be quite happy to help with say > netperf command selection to demonstrate the difference between the two. > > happy benchmrking, > On most modern systems I suspect there will be little to no difference between bonding RX peformance and team performance. If there is any now, I suspect team and bond performance to be similar by the time team has to account for the corner-cases bonding has already resolved. :-) Benchmarks may prove otherwise, but I've yet to see Jiri produce anything. My initial testing doesn't demonstrate any measureable differences with 1Gbps interfaces on a multi-core, multi-socket system.