From: Jason delaFuente (jason.delafuente@GBE.COM)
Date: Tue Mar 23 2004 - 10:58:22 EST
Do you happen to have any additional information or suggestions for good resources that talk about the queue depth on the HBA? Our primary DB machine has 100 16GB LUNS with 4 HBA's connected to an ESS through a switched SAN. We are not seeing any type of excessive wait time nor does iostat show any excessive utilization on any of the underlying hdisk. Even so I would definitely like to know more about this setting. Any suggestions are appreciated.
Thanks!
Jason de la Fuente
>>> RyeR@SCHNEIDER.COM 03/23/04 09:27AM >>>
We have altered our queue depth from the default of 8, but we are using
metavolumes. The queue depth we typically set with metavolumes is 8 * the
number of meta members, which is typically a 4 or 8 member metavolume.
We definitely have seen a difference in performance on metavolumes. The
other item you may want to look at is the "queue depth" of the HBA
[attribute = num_cmd_elems]. By default this is 200, so depending on the
number of luns down the path you may be restricting the performance.
Ralph
Raj Atwal
<atwalrs@YAHOO.CO To: aix-l@Princeton.EDU
M> cc:
Sent by: IBM AIX Fax to:
Discussion List Subject: Queue Depth for EMC
<aix-l@Princeton.
EDU>
03/23/2004 09:05
AM
Please respond to
IBM AIX
Discussion List
I would be interested to know if anybody who is running an EMC disk
subsystem and has altered the queue_depth setting on the hdiskpower
devices?
If so, how did you determine what value to use from the default 8, and how
much of a benefit did you gain in your opinion?
TIA
Raj Atwal
Do you Yahoo!?
Yahoo! Finance Tax Center - File online. File on time.
This archive was generated by hypermail 2.1.7 : Wed Apr 09 2008 - 22:17:44 EDT