From owner-freebsd-performance@FreeBSD.ORG Thu Sep 11 13:13:29 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 43B2A16A4BF for ; Thu, 11 Sep 2003 13:13:29 -0700 (PDT) Received: from mx1.clickcom.com (mx2.clickcom.com [209.198.22.7]) by mx1.FreeBSD.org (Postfix) with ESMTP id E95B343FD7 for ; Thu, 11 Sep 2003 13:13:27 -0700 (PDT) (envelope-from jsmailing@clickcom.com) Received: from aesop (calefaction.clickcom.com [209.198.22.19]) by mx1.clickcom.com (email) with ESMTP id 0192A59EC3; Thu, 11 Sep 2003 16:13:26 -0400 (EDT) From: "John Straiton" To: "'Damian Gerow'" , Date: Thu, 11 Sep 2003 16:08:19 -0400 Message-ID: <002101c378a0$75308380$1916c60a@win2k.clickcom.com> MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit X-Priority: 3 (Normal) X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook, Build 10.0.2627 X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2727.1300 In-Reply-To: <20030911175227.GQ769@sentex.net> Importance: Normal cc: freebsd-performance@freebsd.org Subject: RE: Performance Problems.. Server hardware smoked by $500 box? 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, 11 Sep 2003 20:13:29 -0000 > Hrmmm.... I wonder if 5.0-R is faster than 4-STABLE? The question is really, is it faster than 4.8-S AND 5.1-C. That was the reason we updated the production machine to 5.1 from 4.8... To determine if that was a factor at all. I was hoping that 5.X was just *that much faster* but it would appear that it's not the case unless 5.0-R is *that much faster* than 5.1-C. > > Well. If anything, production should be kicking > development's heiney in this one. You'd think so, at least on the reads. I understand that RAID5 isn't the fastest thing on the block writing, but 99.5% of the traffic to the machine should be requesting pages, not doing disk/database writes other than standard apache logging. > > > - memory speed (PC100 vs. PC133, DDR200 vs. DDR400...) > > Production 512MB PC133 ECC Registered vs Development > 256MB 266Mhz > > DDR > Is the DDR registered at 266MHz? It could be that it's > 532MHz... I don't remember which way the DDR speed rating > goes, I've only just set up my first DDR server yesterday > (didn't trust it just yet). I /think/ it's 266MHz in both > directions, but you'd probably know better than I. I really have no clue. I'm not a hardware guy when it comes to stuff I haven't seen in a server, and I haven't seen anything but various speeds of ECC in the machines I tinker with. > Any network errors on either cards? Have you tried doing > some netperf stuff, to see how the machines handle data transfers? Yeah, netperf scores were in the original email. The development machine topped around 96Mbit/sec and the production one topped like at 87Mbit/sec I think it was. The more interesting numbers were how the development machine nearly *doubled* the production one in NFS scores. > > The servers use Intel integrated cards, vs we have a $7 no-name > > realtek card in the development machine. I made a mistake. The development machine indeed has a $7 realtek in it, but the configured interface is actually a 3c905TX 3Com card. Zero Ierrs, Zero Oerrs, Zero Collisions confirmed 100/full on development Zero Ierrs, Zero Oerrs, Zero Collisions confirmed 100/full on production Of course, the DB server serves files from local but it also has Zero Ierrs, Zero Oerrs, Zero Collisions confirmed 100/full I too have had a periodic problem with auto negotiation on Cisco gear. I wish it was something simple like that here but it'd appear that we're all synced up just fine. > Well, the differences you've pointed out are the RAM and CPU > speed differences. What about other things -- CPU cache > size? What about looking up benchmark differences for the > AMD vs. Intel, and see what they show (you'll probably only > find similar ratings, but if the AMD beats the Intel at > comparitive speeds (1.5GHz), you can bet it'll beat the Intel > at a lower speed. Just make sure you're looking at IO > benchmarks, and not gaming benchmarks.) Well I can't say I know offhand what cache size the Dell PowerEdges use but I'd imagine it's at least equivalent to the AMD chip. I'd say I'd look down that route if it weren't for the next question.. > What about FSB speed -- what does the AMD motherboard run at? > The Intel motherboard? Now I think I just might buy (and agree) that all the difference in the world would be there if I have faster ram, faster cpu AND a faster bus speed. I don't know how to determine that from the DMESG tho' (it didn't seem blantantly obvious to me) so I guess I'd have to try to dig through dell's new site (the site formerly known as the best hardware vendor site when you could put in a ID # and it'd tell you everything exactly for *your* machine instead of the family of machines) to find out what it's running. I could find the MB manual for the development one to look that up if necessary. Would anyone like to concurr that the FSB (et al) could be enough of a difference to explain all this? If so, I'm upgrading my 100Mhz FSB box at home like tomorrow... This development machine just screams in comparison. I've attached the DMESGs from all 3 machines below: John Straiton jks@ clickcom.com Clickcom, Inc 704-365-9970x101 >> Production Web Copyright (c) 1992-2003 The FreeBSD Project. Copyright (c) 1979, 1980, 1983, 1986, 1988, 1989, 1991, 1992, 1993, 1994 The Regents of the University of California. All rights reserved. FreeBSD 5.1-CURRENT #0: Mon Sep 8 12:35:03 EDT 2003 root@:/usr/obj/usr/src/sys/GENERIC Preloaded elf kernel "/boot/kernel/kernel" at 0xc072b000. Preloaded elf module "/boot/kernel/acpi.ko" at 0xc072b1f4. Timecounter "i8254" frequency 1193182 Hz Timecounter "TSC" frequency 927108262 Hz CPU: Intel Pentium III (927.11-MHz 686-class CPU) Origin = "GenuineIntel" Id = 0x68a Stepping = 10 Features=0x387fbff real memory = 1342111744 (1279 MB) avail memory = 1297186816 (1237 MB) Pentium Pro MTRR support enabled npx0: on motherboard npx0: INT 16 interface acpi0: on motherboard acpi0: power button is handled as a fixed feature programming model. Timecounter "ACPI-safe" frequency 3579545 Hz pcibios: BIOS version 2.10 Using $PIR table, 6 entries at 0xc00fc730 acpi_timer0: <32-bit timer at 3.579545MHz> port 0x808-0x80b on acpi0 acpi_cpu0: on acpi0 acpi_cpu1: on acpi0 pcib0: on acpi0 pci0: on pcib0 pcib0: slot 1 INTA is routed to irq 11 pcib0: slot 2 INTA is routed to irq 10 fxp0: port 0xecc0-0xecff mem 0xfe100000-0xfe1fffff,0xfe2f f000-0xfe2fffff irq 11 at device 1.0 on pci0 fxp0: Ethernet address 00:06:5b:3a:48:d5 miibus0: on fxp0 inphy0: on miibus0 inphy0: 10baseT, 10baseT-FDX, 100baseTX, 100baseTX-FDX, auto fxp1: port 0xec80-0xecbf mem 0xfe000000-0xfe0fffff,0xfe2f e000-0xfe2fefff irq 10 at device 2.0 on pci0 fxp1: Ethernet address 00:06:5b:3a:48:d6 miibus1: on fxp1 inphy1: on miibus1 inphy1: 10baseT, 10baseT-FDX, 100baseTX, 100baseTX-FDX, auto pci0: at device 3.0 (no driver attached) isab0: port 0x580-0x58f at device 15.0 on pci0 isa0: on isab0 atapci0: port 0x8b0-0x8bf at device 15.1 on pci0 ata0: at 0x1f0 irq 14 on atapci0 ata1: at 0x170 irq 15 on atapci0 pcib1: on acpi0 pci1: on pcib1 pcib2: on acpi0 pci2: on pcib2 pcib2: slot 5 INTA is routed to irq 5 pcib2: slot 5 INTB is routed to irq 3 ahc0: port 0xdc00-0xdcff mem 0xfeaff000-0xfeafffff irq 5 at device 5 .0 on pci2 aic7899: Ultra160 Wide Channel A, SCSI Id=7, 32/253 SCBs ahc1: port 0xd800-0xd8ff mem 0xfeafe000-0xfeafefff irq 3 at device 5 .1 on pci2 aic7899: Ultra160 Wide Channel B, SCSI Id=7, 32/253 SCBs fdc0: port 0x3f7,0x3f0-0x3f5 irq 6 drq 2 on acpi0 fdc0: FIFO enabled, 8 bytes threshold fd0: <1440-KB 3.5" drive> on fdc0 drive 0 atkbdc0: port 0x64,0x60 irq 1 on acpi0 atkbd0: flags 0x1 irq 1 on atkbdc0 kbd0 at atkbd0 sio0 port 0x3f8-0x3ff irq 4 on acpi0 sio0: type 16550A orm0: