RE: Oracle Book Mal-practice...

  • From: "QuijadaReina, Julio C" <QuijadJC@xxxxxxxxxxxxxxx>
  • To: "rjoralist@xxxxxxxxxxxxxxxxxxxxx" <rjoralist@xxxxxxxxxxxxxxxxxxxxx>, Oracle L <oracle-l@xxxxxxxxxxxxx>
  • Date: Fri, 22 May 2009 16:06:01 -0400

A note about rebuilding indexes. Richard Foote and Jonathan Lewis have some 
excellent articles on the topic too. 
I've noticed that many authors on their books boast about being experts. 
Overtime, some authors have earned reputation of being good or even excellent. 
But even then, everything that is on a book should be first tried and 
researched before considering to use it on production.

Happy Friday and nice weekend!


-----Original Message-----
From: oracle-l-bounce@xxxxxxxxxxxxx [mailto:oracle-l-bounce@xxxxxxxxxxxxx] On 
Behalf Of Rich Jesse
Sent: Friday, May 22, 2009 2:43 PM
To: Oracle L
Subject: RE: Oracle Book Mal-practice...

In a way, the "malpractice" in those kinds of books helps me, at least for
Oracle, which I've worked with for 12+ years now.  Like the idea of
rebuilding indexes, which came up recently.  I believe from past experience
and research that b-tree index rebuilding in Oracle is usually unnecessary. 
But when tasked with "It'll save space, so why not?", I chose to bolster my
belief using other trusted sources (this list's and Tom Kyte's many postings
among them) and use this information to back my assertion.  The
"malpractice" of index rebuilding refreshed my understanding of Oracle's
b-tree indexes.  This is a good thing for me as a "solo" DBA.

OTOH, my <2yrs experience with SQueaL Server is leading me to find all sorts
of questionable information as to how to properly maintain that fine excuse
for an enterprise database.  For example, based on information given on
several websites (including documentation), it seems that SS suffers from
some sort of entropy where time and data contribute to poor performance,
index page corruption, and fragmentation within the table structure causing
disk thrashing, but all only if indexes are not regularly rebuilt or at
least reorg'd.  Really?  Can someone show me how these theorems were proven?

The difference between these is that I consider myself to be a Sr-level in
Oracle and only a Jr in SS land.  But that doesn't mean I should blindly
accept any and all advice given for either.  It just makes it more difficult
to catch all of the BS as a Jr.  And I've missed a few well-flung pieces so

My $.02,

Disclaimer:  I like Fridays.

> The next version of the book should indicate that table and index data be
> separated onto different solid state drives or ram modules when in memory :)
> Seriously these conversations give me such a headache, and usually I am able
> to point to the source being out of date, like a website that hasn't been
> updated since 1997. But this is made even more difficult with current
> sources (books/site/articles etc.) perpetuating these myths.
> On the other hand if everyone learned the proper way to design, develop,
> test, tune and run their systems, the best dba's/developers among us would
> be common and our rate/salaries would reflect that.



Other related posts: