Re: Reviews of Dan Tow's "SQL Tuning"?
From: Shakespeare <whatsin_at_xs4all.nl>
Date: Mon, 12 Jan 2009 22:16:38 +0100
Message-ID: <496bb33a$0$189$e4fe514c_at_news.xs4all.nl>
Charles Hooper schreef:
> On Jan 12, 3:14 pm, hpuxrac <johnbhur..._at_sbcglobal.net> wrote:
>
> I have the second edition (copyright 2001) of Guy Harrison's "Oracle
> SQL High-Performance Tuning". I recall being very impressed with the
> book when I first read it cover to cover, and having just pulled it
> off the shelf to review, there is a great deal of helpful information
> in the book which may be of interest to the OP. But, stay away from
> pages 215-216 of that book. A couple not so nice quotes from that
> section of the book:
> "On Windows NT/Windows 2000, the maximum number of blocks that can be
> read [in a multiblock read] cannot exceed 128K (for 32 bit
> versions)." I trusted that statement until this past summer until I
> saw Oracle 10.2.0.2 on Windows 2000 executing 1MB multiblock reads
> about 8 times faster than 256KB multiblock reads.
> "The size of the Oracle block can have an influence on the efficiency
> of full-table scans. Larger block sizes can often improve scan
> performance." I think that Greg Rahn and Richard Foote have pretty
> well put this one out to pasture.
> "Improving the hit rate in the buffer cache" section title. I think
> that Connor and Jonathan have pretty well demonstrated that this may
> be a lost cause.
>
> I believe that if Guy Harrison updated this book, and fixed those
> items which have later proven to be less than 100% reliable, he would
> have a very useful book.
>
> Charles Hooper
> IT Manager/Oracle DBA
> K&M Machine-Fabricating, Inc.
Date: Mon, 12 Jan 2009 22:16:38 +0100
Message-ID: <496bb33a$0$189$e4fe514c_at_news.xs4all.nl>
Charles Hooper schreef:
> On Jan 12, 3:14 pm, hpuxrac <johnbhur..._at_sbcglobal.net> wrote:
>> Wondering if Guy Harrison will get out a new or updated book at some >> point. I always thought his book was a great starting point for many >> developers new to oracle.
>
> I have the second edition (copyright 2001) of Guy Harrison's "Oracle
> SQL High-Performance Tuning". I recall being very impressed with the
> book when I first read it cover to cover, and having just pulled it
> off the shelf to review, there is a great deal of helpful information
> in the book which may be of interest to the OP. But, stay away from
> pages 215-216 of that book. A couple not so nice quotes from that
> section of the book:
> "On Windows NT/Windows 2000, the maximum number of blocks that can be
> read [in a multiblock read] cannot exceed 128K (for 32 bit
> versions)." I trusted that statement until this past summer until I
> saw Oracle 10.2.0.2 on Windows 2000 executing 1MB multiblock reads
> about 8 times faster than 256KB multiblock reads.
> "The size of the Oracle block can have an influence on the efficiency
> of full-table scans. Larger block sizes can often improve scan
> performance." I think that Greg Rahn and Richard Foote have pretty
> well put this one out to pasture.
> "Improving the hit rate in the buffer cache" section title. I think
> that Connor and Jonathan have pretty well demonstrated that this may
> be a lost cause.
>
> I believe that if Guy Harrison updated this book, and fixed those
> items which have later proven to be less than 100% reliable, he would
> have a very useful book.
>
> Charles Hooper
> IT Manager/Oracle DBA
> K&M Machine-Fabricating, Inc.
I believe he's more into MySQL now. But meanwhile he did a great job by (co)building Quest Spotlight on Oracle. I think his most recent contribution to Oracle Performance Tuning are
http://guyharrison.typepad.com/SystematicOracletuning.pdf
and
http://guyharrison.typepad.com/yadb/2008/03/using-_spin_cou.html
and may I conclude with the quote from is presentation at http://guyharrison.typepad.com/systematicOracletuning.ppt :
"A good buffer cache hit ratio doesn’t help much" (So he learned...)
Shakespeare Received on Mon Jan 12 2009 - 15:16:38 CST