From owner-freebsd-performance@FreeBSD.ORG Mon Jun 16 18:48:40 2003 Return-Path: Delivered-To: freebsd-performance@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 8D84F37B401 for ; Mon, 16 Jun 2003 18:48:40 -0700 (PDT) Received: from irc.dagupan.com (irc.dagupan.com [202.91.161.246]) by mx1.FreeBSD.org (Postfix) with ESMTP id BAE8C43F93 for ; Mon, 16 Jun 2003 18:48:39 -0700 (PDT) (envelope-from francisv-sender-88cbb0@irc.dagupan.com) Received: by irc.dagupan.com (Postfix, from userid 1022) id 8FF5DA7; Tue, 17 Jun 2003 09:48:38 +0800 (PHT) Received: from irc.dagupan.com (localhost [127.0.0.1]) by irc.dagupan.com (Postfix) with ESMTP id 775E3A1 for ; Tue, 17 Jun 2003 09:48:37 +0800 (PHT) Received: from hopper (hopper.dagupan.com [202.91.161.143]) by irc.dagupan.com (tmda-ofmipd) with ESMTP; Tue, 17 Jun 2003 09:48:35 +0800 (PHT) To: Date: Tue, 17 Jun 2003 09:48:33 +0800 X-Mailer: Microsoft Outlook, Build 11.0.4920 Thread-Index: AcM0cjH17PQ0JADoSeypiT/1awP9PgAAFY6Q X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1165 From: Francis Vidal Message-ID: <1055814517.52509.TMDA@irc.dagupan.com> X-Delivery-Agent: TMDA/0.80 (Determine) X-Spam-Status: No, hits=0.3 required=5.5 tests=FROM_HAS_MIXED_NUMS version=2.50-cvs X-Spam-Level: X-Sanitizer: Secured by Bitstop Network Services MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Subject: Xeon with 4GB of RAM running Squid X-BeenThere: freebsd-performance@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list Reply-To: Francis Vidal List-Id: Performance/tuning List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Tue, 17 Jun 2003 01:48:40 -0000 Hi, I have an Intel Xeon machine with 4GB of RAM on FreeBSD 4.7-RELEASE running Squid 2.5-STABLE2. Here's some of the parameters that I've tweaked: /etc/sysctl.conf: vm.v_free_min=131072 vm.v_free_target=262144 vm.f_free_reserved=32768 vm.v_free_severe=65536 kern.ps_showallprocs=0 vfs.vmiodirenable=1 kern.ipc.maxsockbuf=2097152 kern.ipc.somaxconn=8192 kern.ipc.maxsockets=16424 net.inet.tcp.rfc1323=1 net.inet.tcp.delayed_ack=0 net.inet.tcp.sendspace=65535 net.inet.tcp.recvspace=65535 net.inet.udp.recvspace=65535 net.inet.udp.maxdgram=57344 net.local.stream.recvspace=65535 net.local.stream.sendspace=65535 /boot/loader.conf: kern.ipc.maxsockets=32768 kern.ipc.nmbclusters=32768 kern.ipc.nmbufs=131072 kern.ipc.nsfbufs=6656 kern.ipc.shm_use_phys=1 kern.maxfiles=32768 kern.maxproc=8192 kern.maxswzone=33554432 kern.nbuf=16384 kern.ncallout=32768 kern.vm.kmem.size=268435456 kern.vm.pmap.shpgperproc=2048 net.inet.tcp.tcbhashsize=16384 Kernel configuration: machine i386 cpu I686_CPU ident SQUID1 maxusers 128 options MSGMNB=16384 options MSGMNI=41 options MSGSEG=2049 options MSGSSZ=64 options MSGTQL=512 options SHMSEG=16 options SHMMNI=32 options SHMMAX=2097152 options SHMALL=4096 options MAXDSIZ=(2048*1024*1024) options DFLDSIZ=(512*1024*1024) options CPU_ENABLE_SSE options UFS_DIRHASH options KVA_PAGES=512 The cache directory is using software RAID-0 (3 x 40GB 7,200RPM UDMA-6) with softupdates enabled. Any other pointers to utilize this machine's full potential? --- francis a. vidal [bitstop network services] | http://www.bnshosting.net streaming media + web hosting | http://www.bitstop.ph v(02)330-2871,(02)330-2872; f(02)330-2873 | http://www.kuro.ph From owner-freebsd-performance@FreeBSD.ORG Mon Jun 16 18:55:19 2003 Return-Path: Delivered-To: freebsd-performance@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 9461A37B401 for ; Mon, 16 Jun 2003 18:55:19 -0700 (PDT) Received: from mail.cs.ait.ac.th (mail.cs.ait.ac.th [192.41.170.16]) by mx1.FreeBSD.org (Postfix) with ESMTP id CDF0143F3F for ; Mon, 16 Jun 2003 18:55:16 -0700 (PDT) (envelope-from on@cs.ait.ac.th) Received: from banyan.cs.ait.ac.th (on@banyan.cs.ait.ac.th [192.41.170.5]) by mail.cs.ait.ac.th (8.12.3/8.9.3) with ESMTP id h5H1rtv8019862 for ; Tue, 17 Jun 2003 08:55:12 +0700 (ICT) Received: (from on@localhost) by banyan.cs.ait.ac.th (8.8.5/8.8.5) id IAA20205; Tue, 17 Jun 2003 08:55:39 +0700 (ICT) Date: Tue, 17 Jun 2003 08:55:39 +0700 (ICT) Message-Id: <200306170155.IAA20205@banyan.cs.ait.ac.th> X-Authentication-Warning: banyan.cs.ait.ac.th: on set sender to on@banyan.cs.ait.ac.th using -f From: Olivier Nicole Cc: freebsd-performance@freebsd.org In-reply-to: <1055814517.52509.TMDA@irc.dagupan.com> (message from Francis Vidal on Tue, 17 Jun 2003 09:48:33 +0800) References: <1055814517.52509.TMDA@irc.dagupan.com> X-Virus-Scanned: by amavisd-milter (http://amavis.org/) Subject: Re: Xeon with 4GB of RAM running Squid X-BeenThere: freebsd-performance@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list List-Id: Performance/tuning List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Tue, 17 Jun 2003 01:55:19 -0000 > The cache directory is using software RAID-0 (3 x 40GB 7,200RPM UDMA-6) That may be more Squid related than FreeBSD related, but I think I read that Squid is very much able of managing several cache directories, located on several disks, and would end up running faster when let to manage his several cache by itself than through a RAID mechanism. Olivier From owner-freebsd-performance@FreeBSD.ORG Sun Jun 15 01:45:29 2003 Return-Path: Delivered-To: freebsd-performance@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 0E0FD37B401 for ; Sun, 15 Jun 2003 01:45:29 -0700 (PDT) Received: from bluhayz.org (ip68-106-103-50.nv.nv.cox.net [68.106.103.50]) by mx1.FreeBSD.org (Postfix) with ESMTP id 314E343FB1 for ; Sun, 15 Jun 2003 01:45:28 -0700 (PDT) (envelope-from dero@bluhayz.org) Received: from ww2.bluhayz.org (localhost.homeunix.org [127.0.0.1]) by bluhayz.org (8.12.8p1/8.12.8) with ESMTP id h5F9ESNG001130 for ; Sun, 15 Jun 2003 05:14:28 -0400 (EDT) (envelope-from dero@bluhayz.org) From: "agent dero" To: freebsd-performance@freebsd.org Date: Sun, 15 Jun 2003 05:14:28 -0400 Message-Id: <20030615091254.M85497@bluhayz.org> In-Reply-To: <20030614190033.7F0DE37B407@hub.freebsd.org> References: <20030614190033.7F0DE37B407@hub.freebsd.org> X-Mailer: Open WebMail 2.01 20030425 X-OriginatingIP: 172.176.37.186 (dero) MIME-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 X-Mailman-Approved-At: Wed, 18 Jun 2003 10:39:50 -0700 Subject: Re: freebsd-performance Digest, Vol 4, Issue 7 X-BeenThere: freebsd-performance@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list List-Id: Performance/tuning List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sun, 15 Jun 2003 08:45:29 -0000 I have been researching RAID and Software RAID online using FreeBSD's availible documents, and I am wondering what the performance of using a simple RAID 1 capable card over a software RAID 1 configuration? Will the software RAID perform well enough that I could just cut costs of a RAID card? thanks. From owner-freebsd-performance@FreeBSD.ORG Mon Jun 16 18:46:02 2003 Return-Path: Delivered-To: freebsd-performance@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 9037037B401 for ; Mon, 16 Jun 2003 18:46:02 -0700 (PDT) Received: from backup.dagupan.com (mailserver.dagupan.com [202.91.161.131]) by mx1.FreeBSD.org (Postfix) with ESMTP id 4D30243F3F for ; Mon, 16 Jun 2003 18:46:01 -0700 (PDT) (envelope-from francisv@dagupan.com) Received: by mailserver.dagupan.com with Internet Mail Service (5.5.2656.59) id ; Tue, 17 Jun 2003 09:45:55 +0800 Message-ID: <10F29E27A956D511B0940050DA8D86A9084F14@mailserver.dagupan.com> From: francisv@dagupan.com To: freebsd-performance@freebsd.org Date: Tue, 17 Jun 2003 09:45:55 +0800 MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2656.59) Content-Type: text/plain X-Mailman-Approved-At: Wed, 18 Jun 2003 10:39:50 -0700 Subject: Xeon with 4GB of RAM running Squid X-BeenThere: freebsd-performance@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list List-Id: Performance/tuning List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Tue, 17 Jun 2003 01:46:03 -0000 Hi, I have an Intel Xeon machine with 4GB of RAM on FreeBSD 4.7-RELEASE running Squid 2.5-STABLE2. Here's some of the parameters that I've tweaked: /etc/sysctl.conf: vm.v_free_min=131072 vm.v_free_target=262144 vm.f_free_reserved=32768 vm.v_free_severe=65536 kern.ps_showallprocs=0 vfs.vmiodirenable=1 kern.ipc.maxsockbuf=2097152 kern.ipc.somaxconn=8192 kern.ipc.maxsockets=16424 net.inet.tcp.rfc1323=1 net.inet.tcp.delayed_ack=0 net.inet.tcp.sendspace=65535 net.inet.tcp.recvspace=65535 net.inet.udp.recvspace=65535 net.inet.udp.maxdgram=57344 net.local.stream.recvspace=65535 net.local.stream.sendspace=65535 /boot/loader.conf: kern.ipc.maxsockets=32768 kern.ipc.nmbclusters=32768 kern.ipc.nmbufs=131072 kern.ipc.nsfbufs=6656 kern.ipc.shm_use_phys=1 kern.maxfiles=32768 kern.maxproc=8192 kern.maxswzone=33554432 kern.nbuf=16384 kern.ncallout=32768 kern.vm.kmem.size=268435456 kern.vm.pmap.shpgperproc=2048 net.inet.tcp.tcbhashsize=16384 Kernel configuration: machine i386 cpu I686_CPU ident SQUID1 maxusers 128 options MSGMNB=16384 options MSGMNI=41 options MSGSEG=2049 options MSGSSZ=64 options MSGTQL=512 options SHMSEG=16 options SHMMNI=32 options SHMMAX=2097152 options SHMALL=4096 options MAXDSIZ=(2048*1024*1024) options DFLDSIZ=(512*1024*1024) options CPU_ENABLE_SSE options UFS_DIRHASH options KVA_PAGES=512 The cache directory is using software RAID-0 (3 x 40GB 7,200RPM UDMA-6) with softupdates enabled. Any other pointers to utilize this machine's full potential? --- francis a. vidal [bitstop network services] | http://www.bnshosting.net streaming media + web hosting | http://www.bitstop.ph v(02)330-2871,(02)330-2872; f(02)330-2873 | http://www.kuro.ph From owner-freebsd-performance@FreeBSD.ORG Wed Jun 18 10:48:02 2003 Return-Path: Delivered-To: freebsd-performance@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 5299337B404 for ; Wed, 18 Jun 2003 10:48:02 -0700 (PDT) Received: from out004.verizon.net (out004pub.verizon.net [206.46.170.142]) by mx1.FreeBSD.org (Postfix) with ESMTP id 6705243F3F for ; Wed, 18 Jun 2003 10:48:01 -0700 (PDT) (envelope-from cswiger@mac.com) Received: from mac.com ([141.149.47.46]) by out004.verizon.net (InterMail vM.5.01.05.33 201-253-122-126-133-20030313) with ESMTP id <20030618174800.ESJA246.out004.verizon.net@mac.com>; Wed, 18 Jun 2003 12:48:00 -0500 Message-ID: <3EF0A5D0.7040709@mac.com> Date: Wed, 18 Jun 2003 13:48:00 -0400 From: Chuck Swiger Organization: The Courts of Chaos User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.4) Gecko/20030529 X-Accept-Language: en-us, en MIME-Version: 1.0 To: agent dero References: <20030614190033.7F0DE37B407@hub.freebsd.org> <20030615091254.M85497@bluhayz.org> In-Reply-To: <20030615091254.M85497@bluhayz.org> X-Enigmail-Version: 0.76.0.0 X-Enigmail-Supports: pgp-inline, pgp-mime Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit X-Authentication-Info: Submitted using SMTP AUTH at out004.verizon.net from [141.149.47.46] at Wed, 18 Jun 2003 12:47:59 -0500 cc: freebsd-performance@freebsd.org Subject: Re: freebsd-performance Digest, Vol 4, Issue 7 X-BeenThere: freebsd-performance@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list List-Id: Performance/tuning List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 18 Jun 2003 17:48:02 -0000 agent dero wrote: > I have been researching RAID and Software RAID online using FreeBSD's > availible documents, and I am wondering what the performance of using a > simple RAID 1 capable card over a software RAID 1 configuration? Will the > software RAID perform well enough that I could just cut costs of a RAID card? Hardware RAID support can be critical to RAID-5 performance; for RAID-1, the difference between software and hardware probably won't be noticable. [ Yes, this is a generalization that will be false in some circumstances. ] -Chuck From owner-freebsd-performance@FreeBSD.ORG Wed Jun 18 11:53:09 2003 Return-Path: Delivered-To: freebsd-performance@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 0555937B401 for ; Wed, 18 Jun 2003 11:53:09 -0700 (PDT) Received: from mta1.lbl.gov (mta1.lbl.gov [128.3.41.24]) by mx1.FreeBSD.org (Postfix) with ESMTP id 5079743F75 for ; Wed, 18 Jun 2003 11:53:08 -0700 (PDT) (envelope-from j_guojun@lbl.gov) Received: from mta1.lbl.gov (localhost [127.0.0.1]) by mta1.lbl.gov (8.12.9/8.12.9) with ESMTP id h5IIqxIs015752 for ; Wed, 18 Jun 2003 11:53:04 -0700 (PDT) Received: from lbl.gov (gracie.lbl.gov [131.243.2.175]) by mta1.lbl.gov (8.12.9/8.12.9) with ESMTP id h5IIquAs015742; Wed, 18 Jun 2003 11:52:56 -0700 (PDT) Sender: jin@lbl.gov Message-ID: <3EF0B507.2B1B6FDF@lbl.gov> Date: Wed, 18 Jun 2003 11:52:55 -0700 From: "Jin Guojun [DSD]" X-Mailer: Mozilla 4.76 [en] (X11; U; FreeBSD 4.7-RELEASE i386) X-Accept-Language: zh, zh-CN, en MIME-Version: 1.0 To: agent dero References: <20030614190033.7F0DE37B407@hub.freebsd.org> <20030615091254.M85497@bluhayz.org> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit cc: freebsd-performance@freebsd.org Subject: Re: freebsd-performance Digest, Vol 4, Issue 7 X-BeenThere: freebsd-performance@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list List-Id: Performance/tuning List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 18 Jun 2003 18:53:09 -0000 agent dero wrote: > I have been researching RAID and Software RAID online using FreeBSD's > availible documents, and I am wondering what the performance of using a > simple RAID 1 capable card over a software RAID 1 configuration? Will the > software RAID perform well enough that I could just cut costs of a RAID card? > > thanks. Recently, we tested software RAID via CCD and VINUM, and compared them to adaptec 2xxx RAID controller. The performance is the same. The CPU are Xeon 2.8 GHz, SuperMicro MB with ServerWork Chipset. Seagate 3147xxxxLC drives. Problem is the write is slow for all of them, especially when multiple writes. Heard from Adaptect that their 5400 RAID has best performance, but never had one for testing. So, soft RAID is ok to save bucks for hardware. -- ------------ Jin Guojun ----------- v --- j_guojun@lbl.gov --- Distributed Systems Department http://www.itg.lbl.gov/~jin M/S 50B-2239 Ph#:(510) 486-7531 Fax: 486-6363 Lawrence Berkeley National Laboratory, Berkeley, CA 94720 From owner-freebsd-performance@FreeBSD.ORG Wed Jun 18 11:54:58 2003 Return-Path: Delivered-To: freebsd-performance@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 9852C37B401 for ; Wed, 18 Jun 2003 11:54:58 -0700 (PDT) Received: from all.the.wimmins.come.to.loungenet.org (all.the.wimmins.come.to.loungenet.org [64.30.215.225]) by mx1.FreeBSD.org (Postfix) with ESMTP id 07D7643FD7 for ; Wed, 18 Jun 2003 11:54:58 -0700 (PDT) (envelope-from dclements@linkline.com) Received: from all.the.wimmins.come.to.loungenet.org (localhost [127.0.0.1]) id h5IIvdCB002459; Wed, 18 Jun 2003 11:57:39 -0700 (PDT) (envelope-from dclements@linkline.com) Received: (from doug@localhost)h5IIvdJE002458; Wed, 18 Jun 2003 11:57:39 -0700 (PDT) X-Authentication-Warning: all.the.wimmins.come.to.loungenet.org: doug set sender to dclements@linkline.com using -f Date: Wed, 18 Jun 2003 11:57:39 -0700 From: Doug Clements To: "Jin Guojun [DSD]" Message-ID: <20030618185739.GE2215@linkline.com> References: <20030614190033.7F0DE37B407@hub.freebsd.org> <20030615091254.M85497@bluhayz.org> <3EF0B507.2B1B6FDF@lbl.gov> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <3EF0B507.2B1B6FDF@lbl.gov> User-Agent: Mutt/1.4.1i X-Mailman-Approved-At: Wed, 18 Jun 2003 11:55:44 -0700 cc: freebsd-performance@freebsd.org cc: agent dero Subject: Re: freebsd-performance Digest, Vol 4, Issue 7 X-BeenThere: freebsd-performance@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list Reply-To: Doug Clements List-Id: Performance/tuning List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 18 Jun 2003 18:54:58 -0000 On Wed, Jun 18, 2003 at 11:52:55AM -0700, Jin Guojun [DSD] wrote: > Recently, we tested software RAID via CCD and VINUM, and compared > them to adaptec 2xxx RAID controller. The performance is the same. > The CPU are Xeon 2.8 GHz, SuperMicro MB with ServerWork Chipset. > Seagate 3147xxxxLC drives. > > Problem is the write is slow for all of them, especially when multiple writes. > > Heard from Adaptect that their 5400 RAID has best performance, but > never had one for testing. > > So, soft RAID is ok to save bucks for hardware. How was CPU usage during the tests? --Doug From owner-freebsd-performance@FreeBSD.ORG Wed Jun 18 12:24:48 2003 Return-Path: Delivered-To: freebsd-performance@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id C31DD37B401 for ; Wed, 18 Jun 2003 12:24:48 -0700 (PDT) Received: from mta1.lbl.gov (mta1.lbl.gov [128.3.41.24]) by mx1.FreeBSD.org (Postfix) with ESMTP id D8A0143FBD for ; Wed, 18 Jun 2003 12:24:47 -0700 (PDT) (envelope-from j_guojun@lbl.gov) Received: from mta1.lbl.gov (localhost [127.0.0.1]) by mta1.lbl.gov (8.12.9/8.12.9) with ESMTP id h5IJOgIs021124 for ; Wed, 18 Jun 2003 12:24:45 -0700 (PDT) Received: from lbl.gov (gracie.lbl.gov [131.243.2.175]) by mta1.lbl.gov (8.12.9/8.12.9) with ESMTP id h5IJOgAs021118; Wed, 18 Jun 2003 12:24:42 -0700 (PDT) Sender: jin@lbl.gov Message-ID: <3EF0BC78.59821B8C@lbl.gov> Date: Wed, 18 Jun 2003 12:24:41 -0700 From: "Jin Guojun [DSD]" X-Mailer: Mozilla 4.76 [en] (X11; U; FreeBSD 4.7-RELEASE i386) X-Accept-Language: zh, zh-CN, en MIME-Version: 1.0 To: Doug Clements References: <20030614190033.7F0DE37B407@hub.freebsd.org> <20030615091254.M85497@bluhayz.org> <3EF0B507.2B1B6FDF@lbl.gov> <20030618185739.GE2215@linkline.com> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit cc: freebsd-performance@freebsd.org cc: agent dero Subject: Re: freebsd-performance Digest, Vol 4, Issue 7 X-BeenThere: freebsd-performance@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list List-Id: Performance/tuning List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 18 Jun 2003 19:24:49 -0000 Doug Clements wrote: > On Wed, Jun 18, 2003 at 11:52:55AM -0700, Jin Guojun [DSD] wrote: > > Recently, we tested software RAID via CCD and VINUM, and compared > > them to adaptec 2xxx RAID controller. The performance is the same. > > The CPU are Xeon 2.8 GHz, SuperMicro MB with ServerWork Chipset. > > Seagate 3147xxxxLC drives. > > > > Problem is the write is slow for all of them, especially when multiple writes. > > > > Heard from Adaptect that their 5400 RAID has best performance, but > > never had one for testing. > > > > So, soft RAID is ok to save bucks for hardware. > > How was CPU usage during the tests? Very minimum. It was slow on SCSI bus , but I do not have time to diagnose, which can be very time consuming. -Jin From owner-freebsd-performance@FreeBSD.ORG Wed Jun 18 13:05:28 2003 Return-Path: Delivered-To: freebsd-performance@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 284F437B401 for ; Wed, 18 Jun 2003 13:05:28 -0700 (PDT) Received: from otter3.centtech.com (moat3.centtech.com [207.200.51.50]) by mx1.FreeBSD.org (Postfix) with ESMTP id CDAA443F75 for ; Wed, 18 Jun 2003 13:05:26 -0700 (PDT) (envelope-from anderson@centtech.com) Received: from centtech.com (electron.centtech.com [204.177.173.173]) by otter3.centtech.com (8.12.3/8.12.3) with ESMTP id h5IK5P56030682 for ; Wed, 18 Jun 2003 15:05:25 -0500 (CDT) (envelope-from anderson@centtech.com) Message-ID: <3EF0C5FA.302@centtech.com> Date: Wed, 18 Jun 2003 15:05:14 -0500 From: Eric Anderson User-Agent: Mozilla/5.0 (X11; U; Linux i386; en-US; rv:1.0.1) Gecko/20020823 Netscape/7.0 X-Accept-Language: en-us, en MIME-Version: 1.0 To: freebsd-performance@freebsd.org References: <20030614190033.7F0DE37B407@hub.freebsd.org> <20030615091254.M85497@bluhayz.org> <3EF0B507.2B1B6FDF@lbl.gov> <20030618185739.GE2215@linkline.com> <3EF0BC78.59821B8C@lbl.gov> Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit Subject: Re: freebsd-performance Digest, Vol 4, Issue 7 X-BeenThere: freebsd-performance@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list List-Id: Performance/tuning List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 18 Jun 2003 20:05:28 -0000 Jin Guojun [DSD] wrote: > Doug Clements wrote: > > >>On Wed, Jun 18, 2003 at 11:52:55AM -0700, Jin Guojun [DSD] wrote: >> >>>Recently, we tested software RAID via CCD and VINUM, and compared >>>them to adaptec 2xxx RAID controller. The performance is the same. >>>The CPU are Xeon 2.8 GHz, SuperMicro MB with ServerWork Chipset. >>>Seagate 3147xxxxLC drives. >>> >>>Problem is the write is slow for all of them, especially when multiple writes. >>> >>>Heard from Adaptect that their 5400 RAID has best performance, but >>>never had one for testing. >>> >>>So, soft RAID is ok to save bucks for hardware. >> >>How was CPU usage during the tests? > > > Very minimum. > It was slow on SCSI bus , but I do not have time to diagnose, > which can be very time consuming. My tests have shown that a GOOD hardware RAID5 controller can really help you out when it's being slammed (specially over NFS) - I use Dell PERC/2 and PERC/3 controllers (rebranded AMI RAID boards). They run like champs, and I get 40MB/s sustained xfers.. With software RAID, I get less, however it is substantially cheaper, and for price/punch it's hard to beat. It really depends on how much you depend on it, and how much you want to spend on the solution. vinum is one of the sweetest pieces of software ever written to compliment FreeBSD (in my opinion) - it's just one of those tools you can depend on. Eric -- ------------------------------------------------------------------ Eric Anderson Systems Administrator Centaur Technology Attitudes are contagious, is yours worth catching? ------------------------------------------------------------------ From owner-freebsd-performance@FreeBSD.ORG Wed Jun 18 13:26:14 2003 Return-Path: Delivered-To: freebsd-performance@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 2421137B401 for ; Wed, 18 Jun 2003 13:26:14 -0700 (PDT) Received: from mta1.lbl.gov (mta1.lbl.gov [128.3.41.24]) by mx1.FreeBSD.org (Postfix) with ESMTP id 6447F43FBD for ; Wed, 18 Jun 2003 13:26:13 -0700 (PDT) (envelope-from j_guojun@lbl.gov) Received: from mta1.lbl.gov (localhost [127.0.0.1]) by mta1.lbl.gov (8.12.9/8.12.9) with ESMTP id h5IKQ7Is001282 for ; Wed, 18 Jun 2003 13:26:11 -0700 (PDT) Received: from lbl.gov (gracie.lbl.gov [131.243.2.175]) by mta1.lbl.gov (8.12.9/8.12.9) with ESMTP id h5IKQ5As001269; Wed, 18 Jun 2003 13:26:05 -0700 (PDT) Sender: jin@lbl.gov Message-ID: <3EF0CADD.B096A242@lbl.gov> Date: Wed, 18 Jun 2003 13:26:05 -0700 From: "Jin Guojun [DSD]" X-Mailer: Mozilla 4.76 [en] (X11; U; FreeBSD 4.7-RELEASE i386) X-Accept-Language: zh, zh-CN, en MIME-Version: 1.0 To: Eric Anderson References: <20030614190033.7F0DE37B407@hub.freebsd.org> <20030615091254.M85497@bluhayz.org> <3EF0B507.2B1B6FDF@lbl.gov> <3EF0C5FA.302@centtech.com> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit cc: freebsd-performance@freebsd.org Subject: Re: freebsd-performance Digest, Vol 4, Issue 7 X-BeenThere: freebsd-performance@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list List-Id: Performance/tuning List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 18 Jun 2003 20:26:14 -0000 Eric Anderson wrote: > > Very minimum. > > It was slow on SCSI bus , but I do not have time to diagnose, > > which can be very time consuming. > > My tests have shown that a GOOD hardware RAID5 controller can really > help you out when it's being slammed (specially over NFS) - I use Dell > PERC/2 and PERC/3 controllers (rebranded AMI RAID boards). They run > like champs, and I get 40MB/s sustained xfers.. What did you mean sustained xfers ? -- in and out, or just read out. -Jin From owner-freebsd-performance@FreeBSD.ORG Wed Jun 18 13:41:31 2003 Return-Path: Delivered-To: freebsd-performance@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 4B64237B401 for ; Wed, 18 Jun 2003 13:41:31 -0700 (PDT) Received: from silver.he.iki.fi (silver.he.iki.fi [193.64.42.241]) by mx1.FreeBSD.org (Postfix) with ESMTP id 049C143FCB for ; Wed, 18 Jun 2003 13:41:28 -0700 (PDT) (envelope-from pete@he.iki.fi) Received: from PETEX31 (h81.vuokselantie10.fi [193.64.42.129]) by silver.he.iki.fi (8.12.9/8.11.4) with SMTP id h5IKfIcO014986; Wed, 18 Jun 2003 23:41:18 +0300 (EEST) (envelope-from pete@he.iki.fi) Message-ID: <002101c335d9$f6664a20$812a40c1@PETEX31> From: "Petri Helenius" To: "Jin Guojun [DSD]" , "agent dero" References: <20030614190033.7F0DE37B407@hub.freebsd.org><20030615091254.M85497@bluhayz.org> <3EF0B507.2B1B6FDF@lbl.gov> Date: Wed, 18 Jun 2003 23:41:14 +0300 MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 8bit X-Priority: 3 X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook Express 6.00.2800.1158 X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1165 cc: freebsd-performance@freebsd.org Subject: Re: freebsd-performance Digest, Vol 4, Issue 7 X-BeenThere: freebsd-performance@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list List-Id: Performance/tuning List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 18 Jun 2003 20:41:31 -0000 > agent dero wrote: > > > I have been researching RAID and Software RAID online using FreeBSD's > > availible documents, and I am wondering what the performance of using a > > simple RAID 1 capable card over a software RAID 1 configuration? Will the > > software RAID perform well enough that I could just cut costs of a RAID card? > > For mirroring (RAID 1) or mirroring + striping (RAID10 or RAID0+1) the hardware on a RAID card does not really give you a much, just saves you one DMA from the main memory for the other write on the redundant disk. And depending on driver and if you purchased the battery backup on the card, might give you benefits from write ordering. > > Recently, we tested software RAID via CCD and VINUM, and compared > them to adaptec 2xxx RAID controller. The performance is the same. > The CPU are Xeon 2.8 GHz, SuperMicro MB with ServerWork Chipset. > Seagate 3147xxxxLC drives. > > Problem is the write is slow for all of them, especially when multiple writes. > Write shouldn´t be too slow on RAID1. On RAID5 it depends on the implementation, how much overlapping it does and how much overlapping does your test have. > Heard from Adaptect that their 5400 RAID has best performance, but > never had one for testing. > If I understand correctly it has been discontinued. > So, soft RAID is ok to save bucks for hardware. > And currently there are also the reliability issues with drivers like aac (Adaptec 2120S and 2200S, etc.) "no frills" SCSI or ATA controllers run much more reliably and have more user base so the issues get ironed out quicker. Pete From owner-freebsd-performance@FreeBSD.ORG Wed Jun 18 13:44:28 2003 Return-Path: Delivered-To: freebsd-performance@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 5E35537B401 for ; Wed, 18 Jun 2003 13:44:28 -0700 (PDT) Received: from otter3.centtech.com (moat3.centtech.com [207.200.51.50]) by mx1.FreeBSD.org (Postfix) with ESMTP id 939D043F93 for ; Wed, 18 Jun 2003 13:44:25 -0700 (PDT) (envelope-from anderson@centtech.com) Received: from centtech.com (electron.centtech.com [204.177.173.173]) by otter3.centtech.com (8.12.3/8.12.3) with ESMTP id h5IKiO56037104; Wed, 18 Jun 2003 15:44:25 -0500 (CDT) (envelope-from anderson@centtech.com) Message-ID: <3EF0CF1E.9020901@centtech.com> Date: Wed, 18 Jun 2003 15:44:14 -0500 From: Eric Anderson User-Agent: Mozilla/5.0 (X11; U; Linux i386; en-US; rv:1.0.1) Gecko/20020823 Netscape/7.0 X-Accept-Language: en-us, en MIME-Version: 1.0 To: "Jin Guojun [DSD]" References: <20030614190033.7F0DE37B407@hub.freebsd.org> <20030615091254.M85497@bluhayz.org> <3EF0B507.2B1B6FDF@lbl.gov> <3EF0C5FA.302@centtech.com> <3EF0CADD.B096A242@lbl.gov> Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit cc: freebsd-performance@freebsd.org Subject: Re: freebsd-performance Digest, Vol 4, Issue 7 X-BeenThere: freebsd-performance@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list List-Id: Performance/tuning List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 18 Jun 2003 20:44:28 -0000 Jin Guojun [DSD] wrote: > Eric Anderson wrote: > > >>>Very minimum. >>>It was slow on SCSI bus , but I do not have time to diagnose, >>>which can be very time consuming. >> >>My tests have shown that a GOOD hardware RAID5 controller can really >>help you out when it's being slammed (specially over NFS) - I use Dell >>PERC/2 and PERC/3 controllers (rebranded AMI RAID boards). They run >>like champs, and I get 40MB/s sustained xfers.. > > > What did you mean sustained xfers ? -- in and out, or just read out. sustained writes (not burst transfers) and sustained reads.. With hardware RAID's that have cache memory on them (as the PERC's do - mine has 128mb), you can get huge burst speeds (all in memory, so it makes sense). If you have a machine that is writing small files constantly, the cache can significantly help. However, putting a lot of extra memory in the server running a software RAID can do this with similar results. I always suggest piling on the RAM for servers.. Eric -- ------------------------------------------------------------------ Eric Anderson Systems Administrator Centaur Technology Attitudes are contagious, is yours worth catching? ------------------------------------------------------------------ From owner-freebsd-performance@FreeBSD.ORG Wed Jun 18 15:42:16 2003 Return-Path: Delivered-To: freebsd-performance@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 6A10937B401 for ; Wed, 18 Jun 2003 15:42:16 -0700 (PDT) Received: from mta1.lbl.gov (mta1.lbl.gov [128.3.41.24]) by mx1.FreeBSD.org (Postfix) with ESMTP id A9FB543F3F for ; Wed, 18 Jun 2003 15:42:15 -0700 (PDT) (envelope-from j_guojun@lbl.gov) Received: from mta1.lbl.gov (localhost [127.0.0.1]) by mta1.lbl.gov (8.12.9/8.12.9) with ESMTP id h5IMgAIs028632 for ; Wed, 18 Jun 2003 15:42:13 -0700 (PDT) Received: from lbl.gov (gracie.lbl.gov [131.243.2.175]) by mta1.lbl.gov (8.12.9/8.12.9) with ESMTP id h5IMg5As028620; Wed, 18 Jun 2003 15:42:09 -0700 (PDT) Sender: jin@lbl.gov Message-ID: <3EF0EABC.E9D75759@lbl.gov> Date: Wed, 18 Jun 2003 15:42:04 -0700 From: "Jin Guojun [DSD]" X-Mailer: Mozilla 4.76 [en] (X11; U; FreeBSD 4.7-RELEASE i386) X-Accept-Language: zh, zh-CN, en MIME-Version: 1.0 To: Petri Helenius References: <20030614190033.7F0DE37B407@hub.freebsd.org><20030615091254.M85497@bluhayz.org> <3EF0B507.2B1B6FDF@lbl.gov> <002101c335d9$f6664a20$812a40c1@PETEX31> Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 8bit cc: freebsd-performance@freebsd.org cc: agent dero Subject: Re: freebsd-performance Digest, Vol 4, Issue 7 X-BeenThere: freebsd-performance@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list List-Id: Performance/tuning List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 18 Jun 2003 22:42:16 -0000 Petri Helenius wrote: > > > Recently, we tested software RAID via CCD and VINUM, and compared > > them to adaptec 2xxx RAID controller. The performance is the same. > > The CPU are Xeon 2.8 GHz, SuperMicro MB with ServerWork Chipset. > > Seagate 3147xxxxLC drives. > > > > Problem is the write is slow for all of them, especially when multiple writes. > > > Write shouldn´t be too slow on RAID1. On RAID5 it depends on the implementation, > how much overlapping it does and how much overlapping does your test have. The problem is on RAID 1 (mirroring). This may not be the RAID issue, but the SCSI drive and controller combination problem. 10 years ago, Seagate published Elite2 drive, which does not comply with SCSI spec. It took me a quite time to diagnose it. What happened was Elite2 hold SCSi bus for two period of cycles, then release the bus. This killed performance. But this seems to have different problem. At least write to RAID 1 requires twice bus bandwidth for soft RAID 1. So, read should be faster than write. But current read is 67 MB/s, write is 10 MB/s. Something is not working. > > > Heard from Adaptect that their 5400 RAID has best performance, but > > never had one for testing. > > > If I understand correctly it has been discontinued. Yes, but adaptec guy says that go find one if you could because 5400 is good. They told that 2xxxS has lower performance than 5400. > > So, soft RAID is ok to save bucks for hardware. > > > And currently there are also the reliability issues with drivers like aac > (Adaptec 2120S and 2200S, etc.) "no frills" SCSI or ATA controllers > run much more reliably and have more user base so the issues get ironed > out quicker. They are fixing the driver now, hopefully we will get newer one next week. -- ------------ Jin Guojun ----------- v --- j_guojun@lbl.gov --- Distributed Systems Department http://www.itg.lbl.gov/~jin M/S 50B-2239 Ph#:(510) 486-7531 Fax: 486-6363 Lawrence Berkeley National Laboratory, Berkeley, CA 94720 From owner-freebsd-performance@FreeBSD.ORG Wed Jun 18 15:58:33 2003 Return-Path: Delivered-To: freebsd-performance@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 88DD037B401 for ; Wed, 18 Jun 2003 15:58:33 -0700 (PDT) Received: from pop016.verizon.net (pop016pub.verizon.net [206.46.170.173]) by mx1.FreeBSD.org (Postfix) with ESMTP id 97B9043FA3 for ; Wed, 18 Jun 2003 15:58:32 -0700 (PDT) (envelope-from cswiger@mac.com) Received: from mac.com ([141.149.47.46]) by pop016.verizon.net (InterMail vM.5.01.05.33 201-253-122-126-133-20030313) with ESMTP id <20030618225831.DXYF3199.pop016.verizon.net@mac.com> for ; Wed, 18 Jun 2003 17:58:31 -0500 Message-ID: <3EF0EE8D.8040806@mac.com> Date: Wed, 18 Jun 2003 18:58:21 -0400 From: Chuck Swiger Organization: The Courts of Chaos User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.4) Gecko/20030529 X-Accept-Language: en-us, en MIME-Version: 1.0 To: freebsd-performance@freebsd.org References: <20030614190033.7F0DE37B407@hub.freebsd.org> <20030615091254.M85497@bluhayz.org> <3EF0B507.2B1B6FDF@lbl.gov> <002101c335d9$f6664a20$812a40c1@PETEX31> <3EF0EABC.E9D75759@lbl.gov> In-Reply-To: <3EF0EABC.E9D75759@lbl.gov> X-Enigmail-Version: 0.76.0.0 X-Enigmail-Supports: pgp-inline, pgp-mime Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit X-Authentication-Info: Submitted using SMTP AUTH at pop016.verizon.net from [141.149.47.46] at Wed, 18 Jun 2003 17:58:31 -0500 Subject: Re: freebsd-performance Digest, Vol 4, Issue 7 X-BeenThere: freebsd-performance@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list List-Id: Performance/tuning List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 18 Jun 2003 22:58:33 -0000 -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Jin Guojun [DSD] wrote: [ ... ] > But this seems to have different problem. At least write to RAID 1 requires > twice bus bandwidth for soft RAID 1. So, read should be faster than write. > But current read is 67 MB/s, write is 10 MB/s. Something is not working. What is the write performance to a bare hard drive? I'd expect your write performance to be somewhat higher, true, but your drives read at 33 MB/s, and write at 15 MB/s, your write performance is nominal. - -- - -Chuck -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.2 (MingW32) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQE+8O6WHNGCD1IY6vURAjZQAKCG+1TwOtFEHbn4wwU0gWuLLx7RAwCgoEWQ ORk/bmwaFAuncBZlffJ49F4= =6i0F -----END PGP SIGNATURE----- From owner-freebsd-performance@FreeBSD.ORG Wed Jun 18 16:10:03 2003 Return-Path: Delivered-To: freebsd-performance@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id B1D2937B401 for ; Wed, 18 Jun 2003 16:10:03 -0700 (PDT) Received: from mta1.lbl.gov (mta1.lbl.gov [128.3.41.24]) by mx1.FreeBSD.org (Postfix) with ESMTP id 037B843F85 for ; Wed, 18 Jun 2003 16:10:03 -0700 (PDT) (envelope-from j_guojun@lbl.gov) Received: from mta1.lbl.gov (localhost [127.0.0.1]) by mta1.lbl.gov (8.12.9/8.12.9) with ESMTP id h5INA0Iq004256 for ; Wed, 18 Jun 2003 16:10:00 -0700 (PDT) Received: from lbl.gov (gracie.lbl.gov [131.243.2.175]) by mta1.lbl.gov (8.12.9/8.12.9) with ESMTP id h5IN9xAs004253; Wed, 18 Jun 2003 16:09:59 -0700 (PDT) Sender: jin@lbl.gov Message-ID: <3EF0F146.86CE2F33@lbl.gov> Date: Wed, 18 Jun 2003 16:09:58 -0700 From: "Jin Guojun [DSD]" X-Mailer: Mozilla 4.76 [en] (X11; U; FreeBSD 4.7-RELEASE i386) X-Accept-Language: zh, zh-CN, en MIME-Version: 1.0 To: Chuck Swiger References: <20030614190033.7F0DE37B407@hub.freebsd.org> <20030615091254.M85497@bluhayz.org> <3EF0B507.2B1B6FDF@lbl.gov> <3EF0EE8D.8040806@mac.com> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit cc: freebsd-performance@freebsd.org Subject: Re: freebsd-performance Digest, Vol 4, Issue 7 X-BeenThere: freebsd-performance@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list List-Id: Performance/tuning List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 18 Jun 2003 23:10:04 -0000 Chuck Swiger wrote: > -----BEGIN PGP SIGNED MESSAGE----- > Hash: SHA1 > > Jin Guojun [DSD] wrote: > [ ... ] > > But this seems to have different problem. At least write to RAID 1 requires > > twice bus bandwidth for soft RAID 1. So, read should be faster than write. > > But current read is 67 MB/s, write is 10 MB/s. Something is not working. > > What is the write performance to a bare hard drive? 60 MB/s. Multiple writes (3 or 4 drives) are maximized at 147 MB/s. > I'd expect your write performance to be somewhat higher, true, but your drives > read at 33 MB/s, and write at 15 MB/s, your write performance is nominal. If I got 30 MB/s on write, I would not complain at all. -Jin From owner-freebsd-performance@FreeBSD.ORG Wed Jun 18 23:57:48 2003 Return-Path: Delivered-To: freebsd-performance@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 8F5C737B40B; Wed, 18 Jun 2003 23:57:48 -0700 (PDT) Received: from silver.he.iki.fi (silver.he.iki.fi [193.64.42.241]) by mx1.FreeBSD.org (Postfix) with ESMTP id 44D4943FB1; Wed, 18 Jun 2003 23:57:47 -0700 (PDT) (envelope-from pete@he.iki.fi) Received: from PETEX31 (h81.vuokselantie10.fi [193.64.42.129]) by silver.he.iki.fi (8.12.9/8.11.4) with SMTP id h5J6vdcO018893; Thu, 19 Jun 2003 09:57:40 +0300 (EEST) (envelope-from pete@he.iki.fi) Message-ID: <042501c33630$115159a0$812a40c1@PETEX31> From: "Petri Helenius" To: "Jin Guojun [DSD]" References: <20030614190033.7F0DE37B407@hub.freebsd.org><20030615091254.M85497@bluhayz.org> <3EF0B507.2B1B6FDF@lbl.gov> <002101c335d9$f6664a20$812a40c1@PETEX31> <3EF0EABC.E9D75759@lbl.gov> Date: Thu, 19 Jun 2003 09:57:35 +0300 MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 8bit X-Priority: 3 X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook Express 6.00.2800.1158 X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1165 cc: freebsd-performance@freebsd.org cc: agent dero cc: Scott Long Subject: Re: freebsd-performance Digest, Vol 4, Issue 7 X-BeenThere: freebsd-performance@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list List-Id: Performance/tuning List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Thu, 19 Jun 2003 06:57:48 -0000 > But this seems to have different problem. At least write to RAID 1 requires > twice bus bandwidth for soft RAID 1. So, read should be faster than write. > But current read is 67 MB/s, write is 10 MB/s. Something is not working. Might be that whatever you´re using for the write waits for the second write to complete before it issues a write on the first drive again. vinum does this, it does not dispatch both writes at the same time. (I argued about this being "broken" for a while but then gave up, I understand why doing this differently does complicate the code, but at the same time improves performance) > > Yes, but adaptec guy says that go find one if you could because 5400 is good. > They told that 2xxxS has lower performance than 5400. > This would require faith in the fact that the aac driver will get fixed. It should be noted that I have not tested it with 5400 but tests with 2120S have random failures (shutdowns never complete, containerconfig gets read only sporadically, some operations hang for a relatively long time, etc.) so although I would like to use something like the above mentioned cards, it would jump in a dark, even compared with usually downplayed ATA RAID based solutions. > > And currently there are also the reliability issues with drivers like aac > > (Adaptec 2120S and 2200S, etc.) "no frills" SCSI or ATA controllers > > run much more reliably and have more user base so the issues get ironed > > out quicker. > > They are fixing the driver now, hopefully we will get newer one next week. > Scottl fixed a bug related to the aaccli interface about two weeks ago, haven´t heard from him since. It´s good news that the driver is getting actively worked on. (although I keep hearing that his is not exactly getting paid on fixing it) Pete From owner-freebsd-performance@FreeBSD.ORG Thu Jun 19 04:22:15 2003 Return-Path: Delivered-To: freebsd-performance@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 520AF37B401 for ; Thu, 19 Jun 2003 04:22:15 -0700 (PDT) Received: from ns.nojabrsk.ru (ns.nojabrsk.ru [213.141.244.3]) by mx1.FreeBSD.org (Postfix) with ESMTP id 69D6043F93 for ; Thu, 19 Jun 2003 04:22:13 -0700 (PDT) (envelope-from s_porotnikov@nojabrsk.ru) Received: from sergeyco.nojabrsk.ru (sergeyco.nojabrsk.ru [213.141.244.12]) by ns.nojabrsk.ru (8.12.8/8.12.6) with ESMTP id h5JBMAtJ014614 for ; Thu, 19 Jun 2003 17:22:11 +0600 (YEKST) (envelope-from s_porotnikov@nojabrsk.ru) Date: Thu, 19 Jun 2003 17:22:08 +0600 From: s_porotnikov@nojabrsk.ru X-Mailer: The Bat! (v1.61) X-Priority: 3 (Normal) Message-ID: <741132145421.20030619172208@nojabrsk.ru> To: freebsd-performance@freebsd.org MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Subject: freebsd-performance Digest, Vol 4, Issue 9 X-BeenThere: freebsd-performance@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list Reply-To: s_porotnikov@nojabrsk.ru List-Id: Performance/tuning List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Thu, 19 Jun 2003 11:22:15 -0000 I tested software RAID 0 via VINUM on FreeBSD 5.1 three hard drive IBM Ultrastor IC35L073VCD210 connected to onboard Adaptec dual chanel AIC-7899W based controller (64 bit 66MHz PCI), and compared them to Adaptec 2100 RAID controller (32 bit 33MHz PCI). The performance hardware RAID is appeared on 15 % less than VINUM. CPU usage is same. Test utils are rawio and dd with many fork process (8). The CPU are Xeon 2.4 GHz, SuperMicro MB P4DP6Q with Intel E7500 Chipset. mailto:s_porotnikov@nojabrsk.ru From owner-freebsd-performance@FreeBSD.ORG Thu Jun 19 05:22:38 2003 Return-Path: Delivered-To: freebsd-performance@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id B10BD37B401 for ; Thu, 19 Jun 2003 05:22:38 -0700 (PDT) Received: from silver.he.iki.fi (silver.he.iki.fi [193.64.42.241]) by mx1.FreeBSD.org (Postfix) with ESMTP id 612D443FCB for ; Thu, 19 Jun 2003 05:22:37 -0700 (PDT) (envelope-from pete@he.iki.fi) Received: from PETEX31 (h81.vuokselantie10.fi [193.64.42.129]) by silver.he.iki.fi (8.12.9/8.11.4) with SMTP id h5JCMYcO020830; Thu, 19 Jun 2003 15:22:34 +0300 (EEST) (envelope-from pete@he.iki.fi) Message-ID: <05c601c3365d$74938830$812a40c1@PETEX31> From: "Petri Helenius" To: , References: <741132145421.20030619172208@nojabrsk.ru> Date: Thu, 19 Jun 2003 15:22:30 +0300 MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit X-Priority: 3 X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook Express 6.00.2800.1158 X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1165 Subject: Re: freebsd-performance Digest, Vol 4, Issue 9 X-BeenThere: freebsd-performance@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list List-Id: Performance/tuning List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Thu, 19 Jun 2003 12:22:39 -0000 RAID0 is not really RAID since it does not provide any redundancy. Pete ----- Original Message ----- From: To: Sent: Thursday, June 19, 2003 2:22 PM Subject: freebsd-performance Digest, Vol 4, Issue 9 > I tested software RAID 0 via VINUM on FreeBSD 5.1 three hard drive IBM > Ultrastor IC35L073VCD210 connected to onboard Adaptec dual chanel AIC-7899W > based controller (64 bit 66MHz PCI), and compared them to Adaptec 2100 RAID > controller (32 bit 33MHz PCI). The performance hardware RAID is appeared > on 15 % less than VINUM. CPU usage is same. > Test utils are rawio and dd with many fork process (8). > The CPU are Xeon 2.4 GHz, SuperMicro MB P4DP6Q with Intel E7500 Chipset. > > > mailto:s_porotnikov@nojabrsk.ru > > _______________________________________________ > freebsd-performance@freebsd.org mailing list > http://lists.freebsd.org/mailman/listinfo/freebsd-performance > To unsubscribe, send any mail to "freebsd-performance-unsubscribe@freebsd.org" > From owner-freebsd-performance@FreeBSD.ORG Thu Jun 19 05:37:42 2003 Return-Path: Delivered-To: freebsd-performance@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id B5FBD37B401 for ; Thu, 19 Jun 2003 05:37:42 -0700 (PDT) Received: from otter3.centtech.com (moat3.centtech.com [207.200.51.50]) by mx1.FreeBSD.org (Postfix) with ESMTP id BCCA543FA3 for ; Thu, 19 Jun 2003 05:37:41 -0700 (PDT) (envelope-from anderson@centtech.com) Received: from centtech.com (electron.centtech.com [204.177.173.173]) by otter3.centtech.com (8.12.3/8.12.3) with ESMTP id h5JCbe56090886; Thu, 19 Jun 2003 07:37:40 -0500 (CDT) (envelope-from anderson@centtech.com) Message-ID: <3EF1AE85.2080505@centtech.com> Date: Thu, 19 Jun 2003 07:37:25 -0500 From: Eric Anderson User-Agent: Mozilla/5.0 (X11; U; Linux i386; en-US; rv:1.0.1) Gecko/20020823 Netscape/7.0 X-Accept-Language: en-us, en MIME-Version: 1.0 To: s_porotnikov@nojabrsk.ru References: <741132145421.20030619172208@nojabrsk.ru> Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit cc: freebsd-performance@freebsd.org Subject: Re: freebsd-performance Digest, Vol 4, Issue 9 X-BeenThere: freebsd-performance@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list List-Id: Performance/tuning List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Thu, 19 Jun 2003 12:37:43 -0000 s_porotnikov@nojabrsk.ru wrote: > I tested software RAID 0 via VINUM on FreeBSD 5.1 three hard drive IBM > Ultrastor IC35L073VCD210 connected to onboard Adaptec dual chanel AIC-7899W > based controller (64 bit 66MHz PCI), and compared them to Adaptec 2100 RAID > controller (32 bit 33MHz PCI). The performance hardware RAID is appeared > on 15 % less than VINUM. CPU usage is same. > Test utils are rawio and dd with many fork process (8). > The CPU are Xeon 2.4 GHz, SuperMicro MB P4DP6Q with Intel E7500 Chipset. Interestingly enough, I found FreeBSD 4.8 to be faster at local disk writes than FreeBSD 5.1. About 50% faster. Eric -- ------------------------------------------------------------------ Eric Anderson Systems Administrator Centaur Technology Attitudes are contagious, is yours worth catching? ------------------------------------------------------------------ From owner-freebsd-performance@FreeBSD.ORG Thu Jun 19 10:03:07 2003 Return-Path: Delivered-To: freebsd-performance@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 2114037B401; Thu, 19 Jun 2003 10:03:07 -0700 (PDT) Received: from adsl-63-198-35-122.dsl.snfc21.pacbell.net (adsl-63-198-35-122.dsl.snfc21.pacbell.net [63.198.35.122]) by mx1.FreeBSD.org (Postfix) with ESMTP id 3650043FCB; Thu, 19 Jun 2003 10:03:06 -0700 (PDT) (envelope-from j_guojun@lbl.gov) Received: from lbl.gov (localhost.pacbell.net [127.0.0.1]) ESMTP id h5JH6ihc000392; Thu, 19 Jun 2003 10:06:45 -0700 (PDT) (envelope-from j_guojun@lbl.gov) Sender: jin@adsl-63-198-35-122.dsl.snfc21.pacbell.net Message-ID: <3EF1EDA4.CBC4559@lbl.gov> Date: Thu, 19 Jun 2003 10:06:44 -0700 From: "Jin Guojun [NCS]" X-Mailer: Mozilla 4.76 [en] (X11; U; FreeBSD 4.8-RELEASE i386) X-Accept-Language: zh, zh-CN, en-US, en MIME-Version: 1.0 To: Petri Helenius References: <20030614190033.7F0DE37B407@hub.freebsd.org><20030615091254.M85497@bluhayz.org> <3EF0B507.2B1B6FDF@lbl.gov> <002101c335d9$f6664a20$812a40c1@PETEX31> <3EF0EABC.E9D75759@lbl.gov> <042501c33630$115159a0$812a40c1@PETEX31> Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 8bit cc: freebsd-performance@freebsd.org cc: agent dero cc: Scott Long Subject: Re: freebsd-performance Digest, Vol 4, Issue 7 X-BeenThere: freebsd-performance@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list List-Id: Performance/tuning List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Thu, 19 Jun 2003 17:03:07 -0000 Petri Helenius wrote: > > But this seems to have different problem. At least write to RAID 1 requires > > twice bus bandwidth for soft RAID 1. So, read should be faster than write. > > But current read is 67 MB/s, write is 10 MB/s. Something is not working. > > Might be that whatever you´re using for the write waits for the second write > to complete before it issues a write on the first drive again. vinum does this, > it does not dispatch both writes at the same time. (I argued about this > being "broken" for a while but then gave up, I understand why doing > this differently does complicate the code, but at the same time improves > performance) The interesting thing is that the first write is fast every time I rebuilt file system (newfs). The write on soft RAID is about 3# MB/s. After a few large files on the file system, the write speed drops to 11-12 MB/s. Sooner or later, the write speed will stay at 9-10 MB/s. multiple writes will be lower around 8-9 MB/s (aggregated rate). The initial hardware RAID write rate is about 15 MB/s. This may be due to the battery backup write cache does no function correctly. The driver spits follwoing message at the same time: **Monitor** - Battery Fully Changed ... **Monitor** ERROR - Low Battery This may mean that cache is disbaled. -Jin From owner-freebsd-performance@FreeBSD.ORG Thu Jun 19 19:45:35 2003 Return-Path: Delivered-To: freebsd-performance@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 6FAEC37B401 for ; Thu, 19 Jun 2003 19:45:35 -0700 (PDT) Received: from ns.nojabrsk.ru (ns.nojabrsk.ru [213.141.244.3]) by mx1.FreeBSD.org (Postfix) with ESMTP id 8933943F3F for ; Thu, 19 Jun 2003 19:45:33 -0700 (PDT) (envelope-from s_porotnikov@nojabrsk.ru) Received: from sergeyco.nojabrsk.ru (sergeyco.nojabrsk.ru [213.141.244.12]) by ns.nojabrsk.ru (8.12.8/8.12.6) with ESMTP id h5K2jUtJ055320 for ; Fri, 20 Jun 2003 08:45:31 +0600 (YEKST) (envelope-from s_porotnikov@nojabrsk.ru) Date: Fri, 20 Jun 2003 08:45:28 +0600 From: s_porotnikov@nojabrsk.ru X-Mailer: The Bat! (v1.61) X-Priority: 3 (Normal) Message-ID: <1291187545875.20030620084528@nojabrsk.ru> To: freebsd-performance@freebsd.org MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Subject: freebsd-performance Digest, Vol 4, Issue 9 X-BeenThere: freebsd-performance@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list Reply-To: s_porotnikov@nojabrsk.ru List-Id: Performance/tuning List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Fri, 20 Jun 2003 02:45:35 -0000 > Interestingly enough, I found FreeBSD 4.8 to be faster at local disk > writes than FreeBSD 5.1. About 50% faster. On my system results of tests with FreeBSD 4.8 and FreeBSD 5.1 about same. Of course I used UFS2 filesystem on FreeBSD 5.1. > RAID0 is not really RAID since it does not provide any redundancy. I am understand. But I executed performance test. So redundancy may be achieved (in many cases) through elementary backup in simple case or VINUM mirror in other cases. I tested RAID 1+0 - results are identical. I not tested RAID-5 because now hard disks have low cost and using RAID-5 (on practice but not in theory) is unjustified. In many cases for low end solution, used hardware RAID-5 without expensive RAID controllers, having battery cache memory, bring to data loss. -- Best regards, Sergey G. Porotnikov mailto:s_porotnikov@nojabrsk.ru From owner-freebsd-performance@FreeBSD.ORG Sat Jun 21 11:57:50 2003 Return-Path: Delivered-To: freebsd-performance@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 49EB837B401 for ; Sat, 21 Jun 2003 11:57:50 -0700 (PDT) Received: from stoneport.math.uic.edu (stoneport.math.uic.edu [131.193.178.160]) by mx1.FreeBSD.org (Postfix) with SMTP id 9A06C43FAF for ; Sat, 21 Jun 2003 11:57:49 -0700 (PDT) (envelope-from djb-dsn-1056221901.30070@cr.yp.to) Received: (qmail 30071 invoked by uid 1017); 21 Jun 2003 18:58:21 -0000 Date: 21 Jun 2003 18:58:21 -0000 Message-ID: <20030621185821.30070.qmail@cr.yp.to> Automatic-Legal-Notices: See http://cr.yp.to/mailcopyright.html. From: "D. J. Bernstein" To: freebsd-performance@freebsd.org Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline X-Mailman-Approved-At: Sat, 21 Jun 2003 12:58:02 -0700 Subject: ten thousand small processes X-BeenThere: freebsd-performance@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list List-Id: Performance/tuning List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sat, 21 Jun 2003 18:57:50 -0000 FreeBSD 4.8. Test program: malloc(360); malloc(80); malloc(180); malloc(16); malloc(440); sleep(10); _exit(0). Compile statically. The program ends up with 44KB RSS. Where is all that DRAM going? The program also ends up with 168KB VSZ. Where is all that VM going? I don't care much about the 3-page text segment. But I do care about the 39 extra pages of VM, and the 8 extra pages of DRAM. There's no obstacle to having a small program fit into _one_ page per process; two or three can be excused, but 39 is absurd. (Yes, I know that Solaris is worse.) At least 2 pages appear to be wasted by exit(), because it brings in a chunk of stdio, which uses 84 bytes of data and 316 bytes of bss. The libc implementors clearly don't care about 316 bytes of memory, so why don't they make those 316 bytes static? Why doesn't the compiler automatically merge some bss into data when that saves a page? Why can't I omit exit(), manually or automatically, when it's unreachable? Furthermore, malloc() appears to chew up a whole new page of DRAM for each allocation, plus another page---is this counted in VSZ?---for an anonymous mmap. Would it really be that difficult to fit 1076 bytes of requested memory into the 3000-odd bytes available at the end of bss? I sure hope that there's some better explanation for the remaining 32 pages than ``Well, we decided to allocate 131072 bytes of memory for the stack,'' especially when I'm hard-limiting the stack to 4K before exec. ---D. J. Bernstein, Associate Professor, Department of Mathematics, Statistics, and Computer Science, University of Illinois at Chicago From owner-freebsd-performance@FreeBSD.ORG Sat Jun 21 18:19:08 2003 Return-Path: Delivered-To: freebsd-performance@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id BCF8B37B401 for ; Sat, 21 Jun 2003 18:19:08 -0700 (PDT) Received: from mta4.adelphia.net (mta4.adelphia.net [64.8.50.184]) by mx1.FreeBSD.org (Postfix) with ESMTP id D5BFC43F93 for ; Sat, 21 Jun 2003 18:19:05 -0700 (PDT) (envelope-from Support@Netflag.Net) Received: from nfn2.Netflag.Net ([68.69.240.78]) by mta4.adelphia.net (InterMail vM.5.01.05.32 201-253-122-126-132-20030307) with ESMTP id <20030622011904.CDST1347.mta4.adelphia.net@nfn2.Netflag.Net>; Sat, 21 Jun 2003 21:19:04 -0400 Message-Id: <5.2.0.9.2.20030621181039.00ba2a10@pop.dc3.adelphia.net> X-Sender: pedramn@pop.dc3.adelphia.net X-Mailer: QUALCOMM Windows Eudora Version 5.2.0.9 Date: Sat, 21 Jun 2003 18:18:35 -0700 To: "D. J. Bernstein" , freebsd-performance@freebsd.org From: Pedram Nimreezi In-Reply-To: <20030621185821.30070.qmail@cr.yp.to> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii"; format=flowed Subject: Re: ten thousand small processes X-BeenThere: freebsd-performance@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list List-Id: Performance/tuning List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sun, 22 Jun 2003 01:19:09 -0000 I hope that gets corrected to your liking.. I know this is irrelevant, but I just wanted to thank you for tinydns... it makes programming with dns soo much.. what's the technical term? better! generate an ssh key and hop into your service directory and ./add-alias subdomain.domain.com 216.5.5.5.5; make wow... And best of all... I don't need to use a dual processor to run a 500 pound gorilla named BIND. Thank You Doctor Bernstein. Maybe LDAP later on when it's more widely used? You can't get much simpler than you already got it.. svstat /service/tinydns and you get your exact service uptime I have carpal tunnel... I'm sure you could relate anything to save typing is a Good Thing.. Doing that in BIND is hardly similar in triviality. I don't suggest going straight into TinyDNS.. I suggest learning BIND so you really appreciate the craftsmanship of you work and I will continue to phrase it like that to clients and apprentices and the fact that I saw you post on FBSD Hackers ;-) God Bless... At 06:58 PM 6/21/2003 +0000, D. J. Bernstein wrote: >FreeBSD 4.8. Test program: malloc(360); malloc(80); malloc(180); >malloc(16); malloc(440); sleep(10); _exit(0). Compile statically. > >The program ends up with 44KB RSS. Where is all that DRAM going? The >program also ends up with 168KB VSZ. Where is all that VM going? > >I don't care much about the 3-page text segment. But I do care about the >39 extra pages of VM, and the 8 extra pages of DRAM. There's no obstacle >to having a small program fit into _one_ page per process; two or three >can be excused, but 39 is absurd. (Yes, I know that Solaris is worse.) > >At least 2 pages appear to be wasted by exit(), because it brings in a >chunk of stdio, which uses 84 bytes of data and 316 bytes of bss. The >libc implementors clearly don't care about 316 bytes of memory, so why >don't they make those 316 bytes static? Why doesn't the compiler >automatically merge some bss into data when that saves a page? Why can't >I omit exit(), manually or automatically, when it's unreachable? > >Furthermore, malloc() appears to chew up a whole new page of DRAM for >each allocation, plus another page---is this counted in VSZ?---for an >anonymous mmap. Would it really be that difficult to fit 1076 bytes of >requested memory into the 3000-odd bytes available at the end of bss? > >I sure hope that there's some better explanation for the remaining 32 >pages than ``Well, we decided to allocate 131072 bytes of memory for the >stack,'' especially when I'm hard-limiting the stack to 4K before exec. > >---D. J. Bernstein, Associate Professor, Department of Mathematics, >Statistics, and Computer Science, University of Illinois at Chicago >_______________________________________________ >freebsd-performance@freebsd.org mailing list >http://lists.freebsd.org/mailman/listinfo/freebsd-performance >To unsubscribe, send any mail to >"freebsd-performance-unsubscribe@freebsd.org" From owner-freebsd-performance@FreeBSD.ORG Sat Jun 21 23:45:09 2003 Return-Path: Delivered-To: freebsd-performance@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id ECECE37B401 for ; Sat, 21 Jun 2003 23:45:09 -0700 (PDT) Received: from pop016.verizon.net (pop016pub.verizon.net [206.46.170.173]) by mx1.FreeBSD.org (Postfix) with ESMTP id 1F25E43F3F for ; Sat, 21 Jun 2003 23:45:09 -0700 (PDT) (envelope-from cswiger@mac.com) Received: from mac.com ([141.149.47.46]) by pop016.verizon.net (InterMail vM.5.01.05.33 201-253-122-126-133-20030313) with ESMTP id <20030622064508.GPVF3199.pop016.verizon.net@mac.com>; Sun, 22 Jun 2003 01:45:08 -0500 Message-ID: <3EF55072.30104@mac.com> Date: Sun, 22 Jun 2003 02:45:06 -0400 From: Chuck Swiger Organization: The Courts of Chaos User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.4) Gecko/20030612 X-Accept-Language: en-us, en MIME-Version: 1.0 To: freebsd-performance@freebsd.org References: <20030621185821.30070.qmail@cr.yp.to> In-Reply-To: <20030621185821.30070.qmail@cr.yp.to> X-Enigmail-Version: 0.76.0.0 X-Enigmail-Supports: pgp-inline, pgp-mime Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit X-Authentication-Info: Submitted using SMTP AUTH at pop016.verizon.net from [141.149.47.46] at Sun, 22 Jun 2003 01:45:07 -0500 cc: "D. J. Bernstein" Subject: Re: ten thousand small processes X-BeenThere: freebsd-performance@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list List-Id: Performance/tuning List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sun, 22 Jun 2003 06:45:10 -0000 D. J. Bernstein wrote: [ ... ] > I don't care much about the 3-page text segment. But I do care about the > 39 extra pages of VM, and the 8 extra pages of DRAM. There's no obstacle > to having a small program fit into _one_ page per process; two or three > can be excused, but 39 is absurd. (Yes, I know that Solaris is worse.) Indeed-- Solaris insists that all programs be dynamicly linked; Sun claims that staticly linked programs violate the SPARC ABI. [ ... ] > Why doesn't the compiler automatically merge some bss into data when > that saves a page? Most executable object formats seem to prefer to page-align different sections, in VM if not on disk. I don't know ELF well enough to know whether there is an equivalent to the following from /usr/include/mach-o/loader.h, used by NEXTSTEP and MacOS X: * The file type MH_OBJECT is a compact format intended as output of the * assembler and input (and possibly output) of the link editor (the .o * format). All sections are in one unnamed segment with no segment padding. * This format is used as an executable format when the file is so small the * segment padding greatly increases its size. ...but this did (or used to do) the right thing with regard to merging data and bss onto a single page. > Why can't I omit exit(), manually or automatically, when it's unreachable? There's a Usenet thread around the following message-ID: http://www.google.com/groups?selm=aqbgk8%242v4j%241%40shot.codefab.com ...which might be interesting: ]In comp.sys.sun.misc Casper H.S. Dik wrote: ]> "Chuck Swiger" writes: ]>> Under Solaris, /bin/true and /bin/false are shell scripts rather than binary ]>> executables, but a minimal assembly implementation of these two programs ]>> would not need to perform any system calls or invoke any library routines at ]>> all, no? ]> ]> _exit()? ] ]A point. ] ]On the other hand, valid one-line C implementations-- int main() { return 0; ]}-- do not explicitly call exit() or _exit(). They simply return from ]subroutine to a system-provided stub, once called /lib/crt0.o, which was the ]thing that passed control to _main() in the first place. ] ]If Solaris will insist that one dynamicly links crt0.o to have the standard ]system _exit() symbol available, okay. However, nothing in the code above ]requires dynamic linking on systems which provide a static version of _exit. > Furthermore, malloc() appears to chew up a whole new page of DRAM for > each allocation, plus another page---is this counted in VSZ?---for an > anonymous mmap. Would it really be that difficult to fit 1076 bytes of > requested memory into the 3000-odd bytes available at the end of bss? Of course not, at least if you knew at compile time that one could use the equivalent of static arrays, rather than getting memory from a malloc() implementation. For that matter, if you insist upon using dynamic allocation and getting memory at runtime, I suspect that using alloca() would be a win. Hmm, yes: chuck 9426 0.0 0.0 168 44 p0 S+ 2:28AM 0:00.00 foo ...versus... chuck 9430 0.0 0.0 144 28 p0 S+ 2:28AM 0:00.00 foo2 > I sure hope that there's some better explanation for the remaining 32 > pages than ``Well, we decided to allocate 131072 bytes of memory for the > stack,'' especially when I'm hard-limiting the stack to 4K before exec. The malloc manpage has some information that may be relevant: < Reduce the size of the cache by a factor of two. The default cache size is 16 pages. This option can be specified multiple times. [ ... ] EXAMPLES To set a systemwide reduction of cache size, and to dump core whenever a problem occurs: ln -s 'A<' /etc/malloc.conf ...only, setting this option doesn't seem to make any difference to the memory behavior: 183-sec% truss foo readlink("/etc/malloc.conf","<<",63) = 2 (0x2) mmap(0x0,4096,0x3,0x1002,-1,0x0) = 671395840 (0x2804b000) break(0x804d000) = 0 (0x0) break(0x804e000) = 0 (0x0) break(0x804f000) = 0 (0x0) break(0x8050000) = 0 (0x0) break(0x8051000) = 0 (0x0) nanosleep(0xbfbffa68,0xbfbffa60) = 0 (0x0) exit(0x0) process exit, rval = 0 ...nor does setting the malloc flags to ">>" increase the VM usage. But a truss of foo2, using alloca() versus malloc(), shows that we don't get a PAGEZERO section mmap'ed in (!), and we don't invoke brk() to grow the malloc arena: 189-sec% truss foo2 nanosleep(0xbfbff618,0xbfbff610) = 0 (0x0) exit(0x0) process exit, rval = 0 -- -Chuck