Results 1 to 10 of 10

Thread: RAID Not!

  1. #1
    Join Date
    Feb 2001
    Location
    Richmond, VA
    Posts
    9

    Default

    System: G4/500 with System 9.0.4

    Apple ATTO Express Pro UL2 card

    9 gig Atlas 10K LVD drive (Bus 2, ID 2)

    Second 9 gig Atlas 10K LVD drive (Bus 2, ID 4) on the chain.

    I initialized both disks with SoftRaid 2.2.2. I then went to create NEW, Volume Type: Striped-Raid 0, shifted click both drives.
    Under Volume Size: 17366 was shown. After clicking OK, it immediately created a Volume. This is unlike what happened when I created a Raid with my home system, where SoftRAID "built" a Volume (lots of activity with both drives and a Progress Bar appeared, it taking about 5 minutes).
    Now, this new Volume has the SoftRAID icon, except where in SoftRAID the diamond has a little Plus on the icon, on the desktop the vertical bar of that plus is missing, so it's just a dash (call this the weird icon). Don't know what this means if anything.
    Anyway, if I look at the size of the Volume it is 16.94 GB! Obviously it didn't build a Striped Volume (doesn't test like one either, giving same results as when disks are mounted as individual volumes.
    Under the Volumes column in SoftRAID, this new Volume shows as "Raid 0 - 16.96 GB, 2 partitions, SU=128.

    Next, I deleted the Volume and created a single, regular Volume from one of the disks (again, the weird icon). I then selected it and chose "Convert to Stripe."
    This looked like it was going to work as the Prograss Bar came up and the conversion started. But, once it had finished, it started again! The result was a single Volume on the desktop that was about 4 gigs in size! It was like it had created a Striped Volume of half the size of each drive. Unfortunately, this drive tested the same as the drives when not Striped.

    Lastly, now that the drives have been initialized with SoftRAID, Apple's Drive Setup says the drives are not supported and cannot initialize them.

    Obviously something isn't working.

    By the way, this all started with Apples 18gig Atlas 10k and my installed 9 gig (I was going to let 9 gigs go to waste and create a 9 gig Raid). After this didn't work (same problems as above), I swapped out the 18 gig with another 9 gig, thinking same hard drive size might make a difference. It didn't. I also replaced the internal cable and terminator with the cable that came with my Adaptec 39160 card (used at home succesfully with SoftRAID and 2 Atlas 10K II drives). So, it doesn't seem to be a cable problem.

    Help.

  2. #2
    Join Date
    Oct 2000
    Location
    Colorado Springs, CO
    Posts
    936

    Default

    Well, my pair of 9GB Cheetahs striped with SoftRAID have the same icon you describe. The diamond with a little nub on the right side. (like a "-" not a "+"). The volume shows up and tests as a RAID-0 with great results. If you initialized the volumes immediatley prior to creating the stripe, it should take a matter of seconds, because the hard work is done. I have setup a few RAID's and if you do the initialization first it has always been a quick process.

    Looking in SoftRAID, the volume shows on the desktop with that one icon. But in SoftRAID it has the "+" in the icon. The icon we see on the desktop is SoftRAID's logo. The plus (I assume) refers to HFS+ formatting. And the size of the volume is correct. All the formatting and partition mapping eats space. Unformatted drivespace is typically around 5% more than once it is formatted.

    What are you using to benchmark it, what are the settings and what are your results?

    And yes, you need the same size drive for any RAID that is worth its salt. I wouldn't even have tried it otherwise (kudos to you for finding out, though). You went the right route with identical drives and it sounds like all your steps are there. I am really interested in the benchmarks and what you were expecting or seeing on the "other" RAID.

  3. #3
    Join Date
    Feb 2001
    Location
    Richmond, VA
    Posts
    9

    Default

    First, thanks for the response.
    I am using ATTO Express Tools 2.3.1.
    Settings are:
    Max Transfer Size: 512
    Sample Size: 2
    Peak Reads: 69.14, Sust'd Read: 67.95
    Peak Write 138.53 (!), Sust'd Write: 38.01 (!)

    On my Raid at home, 2 Atlas 10K IIs with Adaptec 39160 I get Sust'd read around 200. Can't reacall Sust'd Writes, guess would be around 65.

    Now, if I delete the Raid volume and just mount the two drives as separate volumes, ATTO gives these scores (virtually same for both):

    Same settings:

    Peak Read 71.44, Sust'd Read 70.31 (better then RAID!)
    Peak Write 114.76 (!), Sust'd Write 22.16 (!)

    All tests with System Disk Cache unchecked.

    Thanks

  4. #4
    Join Date
    Feb 2001
    Location
    Richmond, VA
    Posts
    9

    Default

    Update.
    I was able to do a low level format with an older Apple Drive Setup. I then Initialized the drives with SoftRAID and created a new RAID volume. The ATTO benchmarks for the volume now show (same settings as mentioned in earlier post):

    Peak Read: 69.35, Sust'd Read 68.51
    Peak Write: 50.40, Sust'd Write: 42.77

    So, the crazy Peak numbers are gone from the writes. Maybe this is all this setup can do. What is sooo disappointing, is the much cheaper IBM Deskstar 7200 ATA drive in this machine gives the following ATTO Test results:

    Peak Read 61.12, Sust'd Read 59.12
    Peak Write 59.12, Sust'd Write 50.35

  5. #5
    Join Date
    Feb 2001
    Location
    Richmond, VA
    Posts
    9

    Default

    Another item: if I change the Max Transfer Size in the ATTO tool from 512k to 2meg, and then 8meg, the SoftRAID performs much better than the ATA drive. What would be an accurate number (512, 2meg, 4meg) to test with if the volume was to be used as a Photoshop scratch disk? Thanks

  6. #6
    Join Date
    May 2000
    Location
    Cordova, TN
    Posts
    2,428

    Default

    We use 8 MB to get a broader picture.

    Atlas 10k's are not our favorite drives.

    [This message has been edited by Louie (edited 22 February 2001).]

  7. #7
    Join Date
    Feb 2001
    Location
    Richmond, VA
    Posts
    9

    Default

    OK. Do you think these are reasonable results with this card and drives? Thanks.

  8. #8
    Join Date
    May 2000
    Location
    Cordova, TN
    Posts
    2,428

    Default

    I personally have never run Adaptec or Quantum. There may be some comparison numbers on the FTP site or from other members. Are any of your posted numbers from tests with ATTO set at 8 MB?

    My gut feel is that if I had a Sawtooth with its faster PCI bus, faster memory, and faster system bus, than my Power Tower Pro, I would be disappointed. This is based on the fact that I get sustained reads/writes of 77/72 MB/sec using Miles U2Ws, X15 Cheetahs and Granite cables and terminators.

  9. #9
    Join Date
    May 2000
    Location
    wherever I hang my hat
    Posts
    3,575

    Default

    well, for an Ultra2 subsystem, those results are ok. Remember, you are not using an Ultra3 controller with Ultra160 drives.

    I would consider updating the firmware on that card. Also, the latest revision of ATTO Tools is 2.5, I believe.

    I also do not like Quantum drives, but in your case I would boot off the IBM IDE, then point PS etc at the array for scratch space.

  10. #10
    Join Date
    Feb 2001
    Location
    Richmond, VA
    Posts
    9

    Default

    Just want to say THANKS for all the help.

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •