True.  I missed that on the RAID 0 as opposed to 
RAID 1.
 
Also, you would probable be surprised at the number 
of places that use the RAID 6 or RAID ADG in a mirrored 
configuration.  It is not a fast or speedy  solution.  Remember 
that RAID 6 or RAID ADG is the SLOWEST RAID as far as performance is 
concerned.  With RAID 10 you still open yourself up to vulnerability when a 
drive fails on your stripeset which causes your stripeset to fail.  Since 
it is mirrored you are fine unless you have a failure on your mirrored 
stripeset.  Personally, I don't like to use RAID 0 at anytime because it 
has no resiliency when it comes to disk failures.  I would rather have the 
RAID 5 at a minimum for any single disk LUN and then use RAID 1 for additional 
redundancy.  But generally I work with companies that are willing to spend 
the money for that configuration along with having controllers with the maximum 
read/write cache to compensate for the speed.
 
  
  RAID 0 is a striped set, no parity.
RAID 1 is a mirrored 
  set.
So RAID60 would be two RAID 6 arrays striped 
  together.
RAID61 would be two mirrored RAID 6 arrays... I could see 
  maybe why you would strip two RAID 6 arrays to increase performance, but that 
  would be incredibly costly and I would say complete overkill.  If you 
  need redundancy and speed is a high priority, you might as well do RAID10, a 
  stripped set of mirrored drives. However, if you a have a limited number of 
  drives and needed the most storage size by reducing the ratio of parity drives 
  and disk I/O performance isn't too important then RAID 5/6 is your answer. 
  
  On 6/5/07, Phil 
  Thayer <phil.thayer@vitalsite.com> 
  wrote:
  If 
    you do that you need to make sure that the controller will support
RAID 6 
    or RAID ADG.  This is simply a RAID 5 with an additional 
    parity
disk implemented.  This reduces the risk of failure of 
    the entire RAID
if a single disk fails.  The RAID will simply 
    function as if it were a
RAID 5 until the failed disk is physically 
    replaced and the RAID 6 or
RAID ADG is rebuilt.
As an alternative, 
    if you have a controller that does not have RAID 6 or 
RAID ADG, then you 
    can use RAID 5 with a spare disk set aside for use as
a spareset in case 
    of a failure.  This does not eliminate the risk in
case of a 
    single disk failure but it reduces it to the time required to 
rebuild 
    the RAID using the spareset as opposed to the time it takes to
physically 
    replace a drive in a degraded RAID 5. If you suffer a second
disk drive 
    failure during the time that the RAID 5 is rebuilding after 
the first 
    disk drive failure, then you will loose your entire RAID.
The 
    ultimate high availability configuration would be RAID 
    60+.  This
would be two RAID 6 with their own sparesets 
    assigned, mirrored to each 
other.  However, be prepared to 
    loose a larger percentage of your raw
disk drive space.  You 
    will loose the equivalent of:
Two disk drives for each RAID 6 
    used
Two disk drives for each RAID 6 for redundant sparesets 
One raid 
    6 with the mirroring
I really don't expect that you would build 
    something like that for a
home server but I figured I would throw all 
    that out there just in case
you had more money that you know what to do 
    with and want to make sure 
the data on your server is safe from 
    failure.
Phil
> -----Original Message-----
> From: kclug-bounces@kclug.org
> 
    [mailto: 
    kclug-bounces@kclug.org] On Behalf Of Luke-Jr
> Sent: Tuesday, 
    June 05, 2007 9:20 AM
> To: kclug@kclug.org
> Subject: Re: SATA 
    PT2
>
> On Tuesday 05 June 2007 09:11, Phil Thayer wrote: 
    
> > Not to mention that with the recent SATA drive sizes to 
    get
> 1TB of SATA
> > would only take 2 
    drive.  However, if you want to use a
> 
    multi-channel
> > SATA controller with raid you will want to use 
    smaller 
> drives (like 4 x
> > 300 or 8 X 250) so you don't 
    loose too much capacity to parity.
>
> With 8 drives, I'd 
    probably want to make 2 parity for a
> server... As unlikely
> 
    as it is for 2 drives to fail at once, that chance does 
> increase 
    with # of
> drives.
> 
    _______________________________________________
> Kclug mailing 
    list
> Kclug@kclug.org
> http://kclug.org/mailman/listinfo/kclug
>
_______________________________________________
Kclug 
    mailing list
Kclug@kclug.org
http://kclug.org/mailman/listinfo/kclug