[Orca-users] RE: Digest Number 397

[alegrand at wallace.com]

It includes all of the IO on the physical disks in the system once.. 

It does not include the IO from meta devices.  If it did that then
everything 
would end up being over counted.  

consider the simple mirror

d10 -m  d40   d50 
d40 1 1  c1t0d0s7
d50 1 1  c2t0d0s7

the current approach counts the io ocurring on c1to0d0s7, c2t0d0s7, and
ignores the
metadevices

If the metadevice IO was included then you would need to count d10, d40, d50
as well as
the two actual devices.  Getting it to do the right thing (count the IO only
once) would
be a lot more complicated.

Alan LeGrand

-----Original Message-----
From: orca-users at yahoogroups.com [mailto:orca-users at yahoogroups.com]
Sent: Friday, January 25, 2002 6:57 AM
To: orca-users at yahoogroups.com
Subject: [orca-users] Digest Number 397


There is 1 message in this issue.

Topics in this digest:

      1. Question on Disk System wide read/writes per second
           From: "O'Neill, Sean" <sean.oneill at commerceone.com>


________________________________________________________________________
________________________________________________________________________

Message: 1
   Date: Thu, 24 Jan 2002 14:11:08 -0800
   From: "O'Neill, Sean" <sean.oneill at commerceone.com>
Subject: Question on Disk System wide read/writes per second

Curiuos out the code in orcallator.se calculates the system-wide disk
read/write I/Os.

      total_reads    += RAW_disk[i].reads;
      total_writes   += RAW_disk[i].writes;

Is RAW_disk[i] the physical disks - like c3t0whatever?  Does this include
stuff from say SDS including /dev/md/<whatever> ?  I don't think it does.
Just wanted to get clarification.

Sean O'Neill


[This message contained attachments]



________________________________________________________________________
________________________________________________________________________