RE: 1M STRIPE SIZE BEST?

Jared,

If an I/O call size is bigger than the Striping Unit (the size of one chunk), then a single I/O call can motivate participation of 2+ disks in an array. Even if the I/O call size is exactly the same as the striping unit, it can still motivate activity on 2 disks under normal operating conditions. See Hank Tullis's explanation at www.hotsos.com/catalog. Of course, under partial outage conditions, any read to the failed disk and any write to any disk in the array will engage participation from all the living disks in the array.

Cary Millsap
Hotsos Enterprises, Ltd.
http://www.hotsos.com

Upcoming events:

- Hotsos Clinic, Dec 9-11 Honolulu
- Hotsos Clinic 101, Jan 7-9 Knoxville
- Steve Adams's Miracle Master Class, Jan 13-15 Copenhagen
- 2003 Hotsos Symposium, Feb 9-12 Dallas

-----Original Message-----
Still
Sent: Saturday, November 23, 2002 7:14 PM To: Multiple recipients of list ORACLE-L

Maybe I've forgotten something fundamental about the way RAID 5 works, I dunno.

But I'm having a hard time understanding how stripe size is going to determine the number of heads used in an IO.

RAID 5 distributes the data across all spindles, with 1 or more of the spindles for any IO being used for the parity data.

Someone care to enlighten me?

Thanks,

Jared

On Thursday 21 November 2002 06:33, Thomas Day wrote:
> We had a discussion about a year ago here and if I remember correctly
the
> "best" stripe size is a function of number of concurrent users. With
a
> small number of users you want a small stripe size so that each read
i/o
> will utilize the maximum number of heads. It will also tie up the
RAID
> device for the duration of the read. With a large number of users you
want
> a very large stripe size so that a minimum of heads will be used for
each
> read and the RAID device can handle multiple requests at a time.
>
> I don't recall any specific numbers and, as you say, the technology is
> changing so quickly that a discussion from a year ago may not be
pertinent
> to today's environment.
>
>
>
>
> "Post, Ethan"
> <Ethan.Post To: Multiple recipients
of
> list ORACLE-L <ORACLE-L_at_fatcity.com> @ps.net> cc:
> Sent by: root Subject: 1M STRIPE SIZE
BEST?
>
>
> 11/20/2002
> 05:53 PM
> Please
> respond to
> ORACLE-L
>
>
>
>
>
>
> A number of papers recommend a stripe size of 1 M (even for EMC) for
> volumes
> containing data files. I also have the following email from Eyal
Aronoff
> of
> Quest dated Nov 2000. A number of the white papers are more recent.
>
> ============================================================
> The reasons for a larger stripe size on a non-RAID 5 device are:
> 1) Sequential reads are faster if you can take advantege of the read
ahead
> built into the disk caching
> 2) If a 64K read does not start on the first block of the stripe, two
> "spindled" are locked for the duration of the read
>
> However, lately we have been testing some EMC gear and it looks like
EMC
> have optimized both of those for smaller strip size too.
>
> The bottom line - I no longer have an opinion one way or another. The
> undelying technology just changes too rapidly.
>
> Eyal
> ============================================================
>
> Your opinions/comments as far as a "best" practice in setting stripe
sizes
> would be greatly appreciated.
>
> Thanks,
> Ethan
> --
> Please see the official ORACLE-L FAQ: http://www.orafaq.com
> --
> Author: Post, Ethan
> INET: Ethan.Post_at_ps.net
>
> Fat City Network Services -- 858-538-5051 http://www.fatcity.com
> San Diego, California -- Mailing list and web hosting services
> ---------------------------------------------------------------------
> To REMOVE yourself from this mailing list, send an E-Mail message
> to: ListGuru_at_fatcity.com (note EXACT spelling of 'ListGuru') and in
> the message BODY, include a line containing: UNSUB ORACLE-L
> (or the name of mailing list you want to be removed from). You may
> also send the HELP command for other information (like subscribing).

-- 
Please see the official ORACLE-L FAQ: http://www.orafaq.com
-- 
Author: Jared Still
  INET: jkstill_at_cybcon.com

Fat City Network Services    -- 858-538-5051 http://www.fatcity.com
San Diego, California        -- Mailing list and web hosting services
---------------------------------------------------------------------
To REMOVE yourself from this mailing list, send an E-Mail message
to: ListGuru_at_fatcity.com (note EXACT spelling of 'ListGuru') and in
the message BODY, include a line containing: UNSUB ORACLE-L
(or the name of mailing list you want to be removed from).  You may
also send the HELP command for other information (like subscribing).

-- 
Please see the official ORACLE-L FAQ: http://www.orafaq.com
-- 
Author: Cary Millsap
  INET: cary.millsap_at_hotsos.com

Fat City Network Services    -- 858-538-5051 http://www.fatcity.com
San Diego, California        -- Mailing list and web hosting services
---------------------------------------------------------------------
To REMOVE yourself from this mailing list, send an E-Mail message
to: ListGuru_at_fatcity.com (note EXACT spelling of 'ListGuru') and in
the message BODY, include a line containing: UNSUB ORACLE-L
(or the name of mailing list you want to be removed from).  You may
also send the HELP command for other information (like subscribing).