From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from mx1.redhat.com (mx1.redhat.com [172.16.48.31]) by int-mx1.corp.redhat.com (8.11.6/8.11.6) with ESMTP id i9RFRgr26326 for ; Wed, 27 Oct 2004 11:27:42 -0400 Received: from sphinx.mythic-beasts.com (sphinx.mythic-beasts.com [212.69.37.6]) by mx1.redhat.com (8.12.11/8.12.11) with ESMTP id i9RFRf8f013574 for ; Wed, 27 Oct 2004 11:27:41 -0400 Received: from pdw by sphinx.mythic-beasts.com with local (Exim 4.34) id 1CMpi8-000504-6q for linux-lvm@redhat.com; Wed, 27 Oct 2004 16:27:40 +0100 Date: Wed, 27 Oct 2004 16:27:40 +0100 From: Paul Warren Subject: Re: [linux-lvm] High loads when writting to LVM Message-ID: <20041027162740.J24976@sphinx.mythic-beasts.com> References: <417FBB3D.6060002@ist.ucf.edu> Mime-Version: 1.0 Content-Disposition: inline In-Reply-To: <417FBB3D.6060002@ist.ucf.edu>; from jtracy@ist.ucf.edu on Wed, Oct 27, 2004 at 11:14:05AM -0400 Sender: Paul Warren Reply-To: LVM general discussion and development List-Id: LVM general discussion and development List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , List-Id: Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: LVM general discussion and development On Wed, Oct 27, 2004 at 11:14:05AM -0400, Judd Tracy wrote: > I have a system that I am testing out lvm on to get a feeling about > stability and performance. Stability right now seems good, but I am a > little worried about the performance currently I am writing out a large > file ~1TB using dd if=/dev/zero of=test bs=1M and am noticing the load > seams extreamly high "5.21, 5.26, 5.17". Most of the work is being done > in dd but I also notice that kswapd0 is also working fairly heavily. Is > this to be expected with LVM or do i have things setup wrong. Write > performance seems to be around 100MB/s on an XFS filesystem. This may be a result of how load average is calculated. What is the CPU usage when doing these (see top)? On a machine with a large number of processes accessing the disk, it is quite easy for a single, I/O intensive application (such as your dd) to drive the load up. The load average is telling you that you have 5 processes blocked waiting for I/O. Without the dd, the time taken to service each is reduced, and the average number blocked at any given time is therefore lower. On our combined shell/web/email server, an I/O intensive process will result in a load of well over 10, yet responsiveness of the system is not actually that bad, and cpu usage remains low. Paul