[SLE] Thinking about 3ware and Raid 1+0 performance

From: Greg Freemyer (freemyer-ml_at_NorcrossGroup.com)
Date: 08/29/03

  • Next message: Terence McCarthy: "Re: [SLE] Opera and the missing libXM.so.2 - A mystery solved" 
    To: Suse <suse-linux-e@suse.com>
Date: 28 Aug 2003 19:03:59 -0400

    Earlier this week I put a 4-channel 3ware card into a new SuSE box.

    I set it up as Raid 1+0 with default stripe size.

    Then I performed the following on 3 different setups:

    sync; date; dd if=/dev/zero of=target bs=64k count=100000; sync; date

    There was no other activity on the machines and all had the same
    motherboard and disk drives (7200rpm ATAs).

    I tried

    Pure ATA
    3-ware Raid 1
    3-ware Raid 1+0

    To my great surprise I found pure ATA to be the fastest and Raid 1+0 to
    be the slowest by far. (30% slower than Raid 1)

    In particular using iostat -d 10 during the above test I was only
    getting 15,000 blocks per second being written out. I was expecting
    50,000 blocks per second.

    Is there anyone useing 3ware with Raid 1+0 that is getting higher
    throughput than 15,000?

    If so, can you give details of your config.

    Thanks
    Greg 

    -- 
Check the headers for your unsubscription address
For additional commands send e-mail to suse-linux-e-help@suse.com
Also check the archives at http://lists.suse.com
Please read the FAQs: suse-linux-e-faq@suse.com
  • Next message: Terence McCarthy: "Re: [SLE] Opera and the missing libXM.so.2 - A mystery solved"

    Relevant Pages

    • Re: [SLE] Thinking about 3ware and Raid 1+0 performance
      ... I've used a number of these in all sorts of configs and always achieved ... >I set it up as Raid 1+0 with default stripe size. ... >To my great surprise I found pure ATA to be the fastest and Raid 1+0 to ... I was expecting ...
      (SuSE)
    • Boot-Problem (Suse 9.1, externes RAID)
      ... Ein Bekannter hat in seinem Suse 9.1 Server ... Ich konnte remote noch nahezu alle Daten vom internen auf das neue RAID ... Initrd image: /boot/initrd.knoppix ... Added linux-alt ...
      (de.comp.os.unix.linux.misc)
    • Re: New PCI IDE controller or new mainboard - advice needed
      ... Currently '\Program Files' is a junction that points to a folder on a RAID ... Installing to stripe will also allow windows to start faster and all I/O ... When I purchased the two new IDE disks 2 months ago, ...
      (microsoft.public.windowsxp.hardware)
    • Re: Poor raid 1 performance?
      ... I basically assumed a 1-sector stripe ... > Or that the stripe size is lower than available free cache on the RAID ... > delivered to the main system by the RAID controller, ... interface mismatch between drives and host interface. ...
      (comp.sys.ibm.pc.hardware.storage)
    • Re: Oracle IO tuning tips/practices on Windows
      ... I would think that direct attached scsi raid with relatively new ... would read a full stripe. ... to combine the 2 databases into 1 'reporting' database. ...
      (comp.databases.oracle.server)