Top is a I/O good indicator, but too broad. You need to start with the SQL, both 
logical and physical (sequential and scattered reads). 

Always use histograms, this can do great things with minimal effort. With OLTP, there 
is parsing cost component you need to consider. With DW, this is a non-event compared 
with the savings you can gain from sensible CBO decisions. From the logic of the SQL, 
then you can assist the CBO by trying different partitioning schemes, range/hash 
combined, to enable pruning, or at least enable partition-wise joins.

Then you can look at all the file-system/physical volume I/O topics that people have 
written papers about. Sun's site is good place to start.

>>> [EMAIL PROTECTED] 07/24/02 07:44a.m. >>>
Jack - Is your load process CPU-bound (high CPU utilization), or I/O-bound
(high I/O statistics)? My guess is that from waits you want to confirm that
I/O is your high wait statistic, to rule out any other problems that may be
slowing down your load time. Then you may want to make a list of ideas that
may speed your load and benchmark them (run the same load multiple times
under the same circumstances) and see what change or combination of changes
reduces the load time. Once you have a new approach that is yielding better
load performance, you may want to check the wait times again to confirm that
another problem isn't slowing your load.

Dennis Williams
DBA
Lifetouch, Inc.
[EMAIL PROTECTED] 


-----Original Message-----
Sent: Tuesday, July 23, 2002 10:58 AM
To: Multiple recipients of list ORACLE-L


All,

We are tuning a new vital process on our data
warehouse, and it is IO intensive - lots of parallel
direct reads and writes. During our testing we are
driving IO wait to ~60% (per top).

questions:

1) is top a valid measure of IO wait?
2) Is a high io wait an issue to be concerned about?
3) how else can it be accurately measured?
4) How can I link IO wait to what is happening inside
the database?

Thanks,

Jack



--
Please see the official ORACLE-L FAQ: http://www.orafaq.com
--
Author: Binley Lim
  INET: [EMAIL PROTECTED]

Fat City Network Services    -- (858) 538-5051  FAX: (858) 538-5051
San Diego, California        -- Public Internet access / Mailing Lists
--------------------------------------------------------------------
To REMOVE yourself from this mailing list, send an E-Mail message
to: [EMAIL PROTECTED] (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).

Reply via email to