Re: Raid set up

Discussion in 'Computer Information' started by Paul, Mar 15, 2009.

  1. Paul

    Paul Guest

    John wrote:
    > I don't know too much about setting up a Raid array as I've never done
    > it before. However after seeing the following article at toms hardware
    > http://www.tomshardware.com/reviews/short-stroking-hdd,2157.html on
    > short stroking hard drives I am seriously considering setting up a
    > short stroked raid array for my next system.
    >
    > Can anyone tell me how you would ensure against any of the drives in
    > your array from failing if you are running a four drive raid system?
    >
    > I'm also not too sure about what the different raid types mean e.g.
    > raid4 raid5 etc.
    >
    > Cheers,
    >
    > John
    >


    You could start here. This RAID article has a few diagrams.

    http://en.wikipedia.org/wiki/RAID

    RAID0 - Pro - interleaves disks, to enhance bandwidth. 4 disks = 4x the bandwidth
    Buy 4 disks, get the sum of them in capacity. (Remember the 2.2TB limit.)
    Con - no redundancy. A single disk failure means all data lost. The
    more drives used 2,3,4... the higher the odds of a failure occurring.

    RAID1 - Pro - redundant mirror - two identical disks for example
    Con - buy (2) 1TB drives, get 1TB total capacity. This kind of
    redundancy isn't that efficient.
    RAID5 - Pro - efficient redundancy. N+1. Meaning the addition of a parity
    drive, protects against a single drive failure. Buy N+1 drives,
    get N drives worth of capacity. In Windows, the trivial capacity
    limit is 2.2TB. (So 3x1TB drives is the max, without special
    precautions. 3x1TB gives 2TB total capacity in RAID5. 5x500GB drives
    would give 2.0TB capacity etc.)
    Con - Read_Modify_Write for small data changes, reduces speed
    depending on what you're doing. The parity block must be
    recalculated, for it to work. Real hardware RAID5 (RAID
    card with a chip with a heatsink) use that chip to do the
    XOR processing. Real hardware RAID5 also provides a cache
    DIMM slot, and the cache improves the performance.

    RAID0+1 or RAID10 (they're two different things)

    This uses arrays at two levels. For example, you can mirror a pair of drives,
    then interleave with another mirrored pair of drives. Or, you can RAID0 a
    pair of drives, and then mirror them against another RAID0 pair. The interleaving
    part gives more bandwidth. The mirroring part gives redundancy. Both modes
    are protected against a single drive failure. Coverage of a two drive failure,
    differs between the two possible setups. These would also be inconvenient
    to manage. For example, as a novice, you'd be well advised to simulate an
    array failure, while there is no data on your RAID0+1 or RAID10 array,
    then see if you can figure out how to repair it. On motherboards, this
    would typically use 4 drives. I.e. 4x1TB drives gives 2TB capacity, 2x the
    bandwidth, and protection from at least one drive failure.

    You should protect *any* RAID setup, with backups. For example, you could
    built a 4 drive RAID0, using 250GB drives. You get 4x the bandwidth, but seek
    time remains unchanged. You could connect an external 1TB drive in an ESATA
    enclosure, to copy the data at night, and back up the entire array. Then, if
    one drive of the four drive array fails, you have a one-day-old backup to
    use in an emergency. Your external ESATA enclosure should have a fan for cooling,
    to keep the new 1TB drive in good condition. If the backup was sector by
    sector, it might take five hours to run the backup at night.

    The main purpose of redundancy in a RAID array, is to defer maintenance to
    a more convenient time. In a business situation for example, if you mirror
    two drives, and the array is "degraded" by a disk failure at 2PM in the
    afternoon, you can wait until the business is closed for the day, before
    doing maintenance and replacing the hard drive. Since you *always* have
    backups of the computer available, if the entire computer fails, you're
    protected. And entire computers do fail, which is why you have the
    separate backup drive to rely on.

    RAID is not a replacement for doing backups. As long as you
    remember that, you'll be fine. Then, when the array fails
    (and it will), you always have an ace in the hole - just
    reformat, and restore from backups, rather than sending a
    frantic message on USENET, asking for someone to rescue you
    from a mess :)

    If I was building a four drive RAID0, I'd probably purchase a
    fifth drive as a spare. That makes it easier to fix the array
    when one drive fails. Since all the drives have the same capacity,
    it avoids the nuisance of small differences in drive capacity,
    if you attempt to buy another brand as a replacement.

    Now, is it worth doing all of this ? Absolutely not :)
    Use a single drive. Do backups. That is enough headache
    right there. For a novice, RAID is nothing but a nuisance.

    For example, imagine the scenario, that your motherboard just
    failed, and the RAID array was connected to the chipset. Can you
    be sure, that a replacement motherboard will interface to the
    RAID data in the same way ? That is one benefit of using a
    separate RAID card, in that if the motherboard fails, you
    can move the RAID card to another computer. Yes, one Intel
    chipset motherboard may be used to host another Intel RAID array,
    so it can be done, but imagine you're frantic to get your
    data back, and have to find a replacement for your five year
    old motherboard... It is these kinds of scenarios, that make
    RAID not worth it. Personally, I prefer as simple a setup as
    possible, with the widest possible compatibility. This is
    one of the reasons that I've resisted changing to SATA,
    simply because by using all IDE drives, I can take a drive
    from one of my computers, and put it in practically any
    other computer. (Of course, I can fix that with USB
    enclosures, as one way of hiding the interface, so it
    isn't as bad as all that.)

    As for Tomshardware, more so than other sites, they'll write
    any kind of article to generate web site hits. So don't take
    every article they write, as an example of something you should
    run off and do. Their "fill a PC casing with oil for cooling"
    article, is one of the reasons I won't go back to their site.
    Sites like Xbitlabs, take a more reasoned approach to their
    article topics, and they won't do stupid things like fill
    a PC case with oil. Tomshardware is more interested in making
    money, than anything else. It is why they put so much effort
    into "salting" search engines on the web.

    Paul
    Paul, Mar 15, 2009
    #1
    1. Advertising

Want to reply to this thread or ask your own question?

It takes just 2 minutes to sign up (and it's free!). Just click the sign up button to choose a username and then you can ask your own questions on the forum.
Similar Threads
  1. Mod
    Replies:
    0
    Views:
    782
  2. Replies:
    0
    Views:
    755
  3. SATA - Raid and Non Raid Question

    , Jan 10, 2007, in forum: Computer Support
    Replies:
    1
    Views:
    775
  4. =?Utf-8?B?VGhlb3JldGljYWxseQ==?=

    Does x64 require a SATA RAID Driver to install non-RAID SATA Drive

    =?Utf-8?B?VGhlb3JldGljYWxseQ==?=, Jul 15, 2005, in forum: Windows 64bit
    Replies:
    6
    Views:
    876
    Charlie Russel - MVP
    Jul 18, 2005
  5. Lawrence D'Oliveiro

    RAID 5 Down, RAID 6 To Go

    Lawrence D'Oliveiro, Feb 23, 2010, in forum: NZ Computing
    Replies:
    0
    Views:
    423
    Lawrence D'Oliveiro
    Feb 23, 2010
Loading...

Share This Page