From mboxrd@z Thu Jan 1 00:00:00 1970 From: Konrad Rzeszutek Wilk Subject: Re: [Xen-devel] [PATCH 0/4] Implement persistent grant in xen-netfront/netback Date: Fri, 16 Nov 2012 10:23:25 -0500 Message-ID: <20121116152325.GG22320@phenom.dumpdata.com> References: <1352962987-541-1-git-send-email-annie.li@oracle.com> <20121115074057.GO8912@reaktio.net> <50A4AA06.8080900@oracle.com> <50A4CA51.8080208@citrix.com> <1352978106.3499.101.camel@zakaz.uk.xensource.com> <20121115182928.GB22320@phenom.dumpdata.com> <1353006667.26243.6.camel@dagon.hellion.org.uk> Mime-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: QUOTED-PRINTABLE Cc: Roger Pau Monne , ANNIE LI , Pasi =?iso-8859-1?Q?K=E4rkk=E4inen?= , "netdev@vger.kernel.org" , "xen-devel@lists.xensource.com" To: Ian Campbell Return-path: Received: from userp1040.oracle.com ([156.151.31.81]:19572 "EHLO userp1040.oracle.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751862Ab2KPPXo convert rfc822-to-8bit (ORCPT ); Fri, 16 Nov 2012 10:23:44 -0500 Content-Disposition: inline In-Reply-To: <1353006667.26243.6.camel@dagon.hellion.org.uk> Sender: netdev-owner@vger.kernel.org List-ID: On Thu, Nov 15, 2012 at 07:11:07PM +0000, Ian Campbell wrote: > On Thu, 2012-11-15 at 18:29 +0000, Konrad Rzeszutek Wilk wrote: > > On Thu, Nov 15, 2012 at 11:15:06AM +0000, Ian Campbell wrote: > > > On Thu, 2012-11-15 at 10:56 +0000, Roger Pau Monne wrote: > > > > On 15/11/12 09:38, ANNIE LI wrote: > > > > >=20 > > > > >=20 > > > > > On 2012-11-15 15:40, Pasi K=E4rkk=E4inen wrote: > > > > >> Hello, > > > > >> > > > > >> On Thu, Nov 15, 2012 at 03:03:07PM +0800, Annie Li wrote: > > > > >>> This patch implements persistent grants for xen-netfront/ne= tback. This > > > > >>> mechanism maintains page pools in netback/netfront, these p= age pools is used to > > > > >>> save grant pages which are mapped. This way improve perform= ance which is wasted > > > > >>> when doing grant operations. > > > > >>> > > > > >>> Current netback/netfront does map/unmap grant operations fr= equently when > > > > >>> transmitting/receiving packets, and grant operations costs = much cpu clock. In > > > > >>> this patch, netfront/netback maps grant pages when needed a= nd then saves them > > > > >>> into a page pool for future use. All these pages will be un= mapped when > > > > >>> removing/releasing the net device. > > > > >>> > > > > >> Do you have performance numbers available already? with/with= out persistent grants? > > > > > I have some simple netperf/netserver test result with/without= persistent=20 > > > > > grants, > > > > >=20 > > > > > Following is result of with persistent grant patch, > > > > >=20 > > > > > Guests, Sum, Avg, Min, Max > > > > > 1, 15106.4, 15106.4, 15106.36, 15106.36 > > > > > 2, 13052.7, 6526.34, 6261.81, 6790.86 > > > > > 3, 12675.1, 6337.53, 6220.24, 6454.83 > > > > > 4, 13194, 6596.98, 6274.70, 6919.25 > > > > >=20 > > > > >=20 > > > > > Following are result of without persistent patch > > > > >=20 > > > > > Guests, Sum, Avg, Min, Max > > > > > 1, 10864.1, 10864.1, 10864.10, 10864.10 > > > > > 2, 10898.5, 5449.24, 4862.08, 6036.40 > > > > > 3, 10734.5, 5367.26, 5261.43, 5473.08 > > > > > 4, 10924, 5461.99, 5314.84, 5609.14 > > > >=20 > > > > In the block case, performance improvement is seen when using a= large > > > > number of guests, could you perform the same benchmark increasi= ng the > > > > number of guests to 15? > > >=20 > > > It would also be nice to see some analysis of the numbers which j= ustify > > > why this change is a good one without every reviewer having to ev= aluate > > > the raw data themselves. In fact this should really be part of th= e > > > commit message. > >=20 > > You mean like a nice graph, eh? >=20 > Together with an analysis of what it means and why it is a good thing= , > yes. OK, lets put that on the TODO list for the next posting. In the meantim= e - it sounds like you (the maintainer) are happy with the direction this i= s going. The other things we want to do _after_ these patches is to look at the = Wei Liu patches and try to address the different reviewers comments. The neat thing about them is that they have a concept of a page pool sy= stem. And with persistent pages in both blkback and netback this gets more ex= citing. >=20 > Ian. >=20 > >=20 > > I will run these patches on my 32GB box and see if I can give you > > a nice PDF/jpg. > >=20 > > >=20 > > > Ian. > > >=20 >=20