[rescue] Re: [SunHELP] Help with RAID

Wed Jun 4 23:18:07 CDT 2003

>Date: Wed, 4 Jun 2003 15:50:54 -0400
>From: Phil Stracchino <alaric at caerllewys.net>
>To: The SunHELP List <sunhelp at sunhelp.org>
>Mail-Followup-To: The SunHELP List <sunhelp at sunhelp.org>, rescue at sunhelp.org
>X-ICBM: 35.6880N 77.4375W
>X-PGP-Fingerprint: 2105 C6FC 945D 2A7A 0738  9BB8 D037 CE8E EFA1 3249
>X-PGP-Key-FTP-URL: ftp://ftp.babcom.com/pub/pgpkeys/alaric.asc
>X-PGP-Key-HTTP-URL: http://www.babcom.com/alaric/pgp.html
>X-UCE-Policy: No unsolicited commercial email is accepted at this site.  All 
senders of UCE will be immediately and permanently blocked.
>User-Agent: Mutt/1.5.4i
>Cc: rescue at sunhelp.org
>Subject: [rescue] Re: [SunHELP] Help with RAID
>
>On Wed, Jun 04, 2003 at 03:17:32PM -0400, robert brooke gravitt wrote:
>> Ok,
>> 
>> I've looked through the Sun doc on DiskSuite, but I'm still a little 
>> clueless on what to do.
>> 
>> I've got a Sun SPARCStorage array with 23 Disks in it ( 23 x 1 Gig.) 
>> What i'd like to do is create a striped, mirrored volume for storage. 
>> I'd like to make 2 10 Gig mirrors to mount and use for a CVS root. Any 
>> help will be appreciated. I looked through a tutorial for mirroring 
>> your root disk, but I just don't understand how to concatenate the 
>> disks and then mirror them.
>
>Fortunately, what you're trying to do is something in which Solaris
>Disksuite knows to Do The Right Thing.  You see, the way most people
>think of doing this is to create two stripes and then mirror them, but
>the correct way is to create N mirrors and stripe them.  Disksuite knows
>this, so when you ask for the former, it's bright enough to do the
>latter.  (This, at least, is my understanding as explained to me by a
>technical wizard in Sun's High Performance Computing group.)

Ummm... everywhere I've ever done this, you create either two stripes
which you then mirror, or two concats and then mirror.... I've never
heard of someone concatenating a bunch of little mirrors...

>
>
>Here's an example of the metainit commands to do it:
>
>metainit d11 1 10 c1t{0,1,2,3,4,5,6,7,8,9}d0s0 -i 128k
>metainit d12 1 10 c2t{0,1,2,3,4,5,6,7,8,9}d0s0 -i 128k
>metainit d10 -m d11
>metattach d10 d12

This is creating a mirror of two stripes !

>
>Oh yeah, don't use the default 16k interlace factor.  It's too small for
>optimal efficiency.  You'll get better performance with the 128k
>interlace noted above.

depends on what you are storing and what your disk access is like.
Generally a good recommendation is to go by the avg size of a disk
cylinder or less (to avoid head movement during the access... let the
head move between the stripe accesses).

>
>Although this looks like two stripes mirrored to each other, internally
>Disksuite implements this as a stripe of ten individual two-disk mirrors.

Say what ?  Is this in the docs somewhere.... I've never heard of this.
It also runs contradictory to what you are setting up.... you are first
creating the ODS stripes, then creating an ODS mirror.  Internally if
it treats it as something other than this would be quite odd to handle.

Are you sure about this ?  I'd like to know more about it and the source,
becasue I worked for Sun, have friends that work for Sun, have used Disk Suite
since the SunOS 4.X days and I have never heard of this.

>
>Why does it do this? Redundancy.  With two stripes mirrored to each
>other, if one single disk in each stripe fails, you're offline and your
>data is history.  With a stripe of ten individual submirrors, you can
>survive failures of half your total disks without data loss, so long as
>you don't lose both disks of any mirrored pair, and the performance is
>identical to doing it the obvious, easy (but wrong) way.

OK... this makes sense... but I'd still think you need to create the
10 submirrors which you would then stripe.  I doubt ODS is going to do
it automagically as such hiding the fact that it has done so...

>
>Calculating the statistical odds of losing both disks of one mirror vs.
>losing one disk from each stripe is left as an exercise for the reader.

An excersize not really worth doing... it is obvious that there is less
chance of losing the correct minimum two disks to take out the striped
submirrors than one disk each of the mirrored stripes....

I question the chances of either happening before the first disk is
replaced (at least in a production environment).

If you have the CPU... you'll get more space by going with RAID 5,
but your I/O throughput will go down.  Another instance of space vs speed.

-- Curt

>
>
>-- 
> .*********  Fight Back!  It may not be just YOUR life at risk.  *********.
> : phil stracchino : unix ronin : renaissance man : mystic zen biker geek :
> :  alaric at caerllewys.net : alaric-ruthven at earthlink.net : phil at latt.net  :
> :   2000 CBR929RR, 1991 VFR750F3 (foully murdered), 1986 VF500F (sold)   :
> :    Linux Now!   ...Because friends don't let friends use Microsoft.    :
>_______________________________________________
>rescue list - http://www.sunhelp.org/mailman/listinfo/rescue

Curtis Wilbar
Hawk Mountain Networks
rescue at hawkmountain.net

My e-mail is protected against viruses and spam by MailGuardian
                  http://www.mailguardian.net
          Top notch protection at unbelievable prices