Hi Jonathan
Would like to have Tests done for BOTH Small Big DB_BLOCK_SIZE ,
if possible , as mentioned below
Thanks
-Original Message-
Sent: Wednesday, December 18, 2002 4:50 PM
To: Multiple recipients of list ORACLE-L
I think there are too many generic arguments
available for
Sorry,
It was a rhetorical question.
Detailed results come under the heading
of company confidential - generic results
come under the heading of repetition.
You just have to list the set of point (see my
earlier point) about why smaller or larger blocks
MIGHT make a difference that you could
And just to add my bit...the possible performance
gains with various blocksizes under various conditions
all assume that your underlying OS layer provides good
support for what you want to do.
Whilst hardly being definitive, I remember (a few
years ago now) a few colleagues of mine had access to
Title: RE: db block Size for Indexes Tablespaces in 9.2 ?
Tim,
it is no rant at all, at-least folks like (yours truly) can learn different aspects that must be considered in tuning.
Thanks
Raj
__
Rajendra Jamadagni MIS, ESPN Inc
Please pardon the forthcoming rant...
RANT
Block size is one of the last things to consider in tuning indexes. Yes,
it's a knob that can be twiddled, but it should be one of those
out-of-the-way knobs that hardly ever get touched. It's much the same as
adjusting DB_BLOCK_BUFFERS in order to
I think there are too many generic arguments
available for picking the 'right' block size for
your indexes.
The one that is most appropriate is likely to
depend on the nature of the activity (load
vs. query), nature of the index (unique,
nearly unique, far from unique), data clustering,
order of
Title: RE: db block Size for Indexes Tablespaces in 9.2 ?
We use 8k for data and 8k for indexes in our OLTP system ...
Raj
__
Rajendra Jamadagni MIS, ESPN Inc.
Rajendra dot Jamadagni at ESPN dot com
Any opinion expressed here is personal
nathan Lewis" [EMAIL PROTECTED]
To: "Multiple recipients of list ORACLE-L"
[EMAIL PROTECTED]
Sent: Wednesday, December 18, 2002 6:19
AM
Subject: Re: db block Size for Indexes Tablespaces
in 9.2 ?
I think there are too many generic arguments
available for picking the 'right' blo
What is the Best Practice for DB_BLOCK_SIZE for Index Tablespaces Undergoing High
Volume of Insert , Update , Selects ? Why ?
Any Docs , Links for the Same ?
NOTE - This is for a Hybrid Application Benchmark undergoing 2000 Banking Transactions
per second
approx.
OLTP Transaction sample =
Title: RE: db block Size for Indexes Tablespaces in 9.2 ?
Can you specify db_block_sizes at the tablespace level in 9?
*blink*
-Original Message-
From: VIVEK_SHARMA [mailto:[EMAIL PROTECTED]]
Sent: Tuesday, December 17, 2002 8:29 AM
To: Multiple recipients of list ORACLE-L
I can't point to a paper or article on that, but based on tests I have done,
the block size of index tablespaces for a high volume transaction system
should be lower than the table block sizes. 2000 tps with 25 changes each
does not sound like a really huge transaction rate. You could set the
Hi Arup , List
Your point is Correct about High buffer busy wait Contention During Large OLTP
Insert /Updates.
High buffer busy wait on Corresponding INDEX during INSERT Operations was Observed
during our previous benchmark which overcame by Converting to REVERSE Index as the
Field Value
was
12 matches
Mail list logo