From owner-freebsd-questions@FreeBSD.ORG Fri Jul 22 21:10:38 2011 Return-Path: Delivered-To: questions@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:4f8:fff6::34]) by hub.freebsd.org (Postfix) with ESMTP id 91A1F1065670 for ; Fri, 22 Jul 2011 21:10:38 +0000 (UTC) (envelope-from nomadlogic@gmail.com) Received: from mail-yx0-f182.google.com (mail-yx0-f182.google.com [209.85.213.182]) by mx1.freebsd.org (Postfix) with ESMTP id 519828FC13 for ; Fri, 22 Jul 2011 21:10:37 +0000 (UTC) Received: by yxl31 with SMTP id 31so1832377yxl.13 for ; Fri, 22 Jul 2011 14:10:37 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=gamma; h=mime-version:in-reply-to:references:date:message-id:subject:from:to :cc:content-type; bh=c812IfTHfdd45M5DJ+/A1zjuXmfoIQXD+MlG6UYmOkY=; b=PLTAZcU1Hec6DLTA5EomhHBnpz0eQTZX2YM06EoCsk+sWErmS5oR2QxUTs/gDpXhLp 48wv6fdvPpTTw+c89AlegHNse9m3iQQGDo1VXEVOD3p1X0a8+C6MmsiLw7w6qF8bQuMe 5OzjaYBGAADIgHF8TZ2axce20f1DQDTumeNGs= MIME-Version: 1.0 Received: by 10.236.144.196 with SMTP id n44mr2970146yhj.229.1311367294723; Fri, 22 Jul 2011 13:41:34 -0700 (PDT) Received: by 10.236.60.227 with HTTP; Fri, 22 Jul 2011 13:41:34 -0700 (PDT) In-Reply-To: <20110722202635.GA57111@bewilderbeast.blackhelicopters.org> References: <20110721141011.GB47190@bewilderbeast.blackhelicopters.org> <20110721151210.GA47663@bewilderbeast.blackhelicopters.org> <4E29BC1F.6060103@sentex.net> <20110722190821.GA55040@bewilderbeast.blackhelicopters.org> <4E29D37E.8070908@sentex.net> <20110722201034.GA56830@bewilderbeast.blackhelicopters.org> <4E29DA4F.9010006@sentex.net> <20110722202635.GA57111@bewilderbeast.blackhelicopters.org> Date: Fri, 22 Jul 2011 13:41:34 -0700 Message-ID: From: pete wright To: "Michael W. Lucas" Content-Type: text/plain; charset=ISO-8859-1 Cc: questions@freebsd.org, Mike Tancsa Subject: Re: em0 NIC slow on 8.2-p1 amd64? X-BeenThere: freebsd-questions@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: User questions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Fri, 22 Jul 2011 21:10:38 -0000 On Fri, Jul 22, 2011 at 1:26 PM, Michael W. Lucas wrote: > On Fri, Jul 22, 2011 at 04:15:11PM -0400, Mike Tancsa wrote: >> On 7/22/2011 4:10 PM, Michael W. Lucas wrote: > Will applications such as NFS cut bandwith usage that much? > I have seen similar performance degradations with NFS in the past. I have seem cases where throughput is hurt due to frequent getattr() calls by the NFS client (esp noticable on Linux hosts traversing large namespaces fwiw). Some possible workarounds/tweaks: 1) increase rsize/wsize (32k for larger files for example) of client mount 2) if performance is only requirement UDP will increase performance versus TCP with obvious downside of using UDP :) 3) jumbo frames (MTU=9000) should help in most cases if available I've also done a bit of testing with NFSv4 - and I find performance here can be a bit better than v3 due to better attribute caching (decreasing amount of getattr() calls when traversing filesystems) and other interesting bigs v4 has. Granted moving from v3 to v4 is not trivial... just my two bits :) -pete -- pete wright www.nycbug.org