Re: Oracle 911 Article
- From: "Richard Foote" <richard.foote@xxxxxxxxxxx>
- To: <oracle-l@xxxxxxxxxxxxx>
- Date: Fri, 5 Mar 2004 23:49:43 +1000
Oracle 911 ArticleI'm not sure if it was a technical article or an attempt
at a resume !!
It certainly has it's amusing moments. Can you imagine the scene. A patient
lies critically ill on the surgery table when the doctor suddenly utters
"Shit, some bastard has just gone and deleted all the statistics from our
"Where To Find Bodily Parts (WTFBP)" database, we're doomed". "Oh No"
exclaims the nurse, "quick, someone fetch Don" !!
Strangely what he calls "Unconventional Methods" I call "Conventional and
Elementary Methods" in tuning trivial problems. Problem: missing indexes
Solution: create missing indexes, Problem: missing statistics Solution add
missing statistics, etc doesn't sound particularly "unconventional" to me
(although if the site is migrating to CBO, I would prefer to just go back to
the RBO as a safer emergency solution, if not migrating then what the hell
happened to all the stats bar one table)
His "Contrary to the pontifications of theoreticians and ivory-tower
academics, there are many silver bullets for Oracle performance tuning"
comments could possible by aimed at yours truly, for which I would be deeply
flattered for the complement. However as I work full-time as a DBA dealing
with real databases in real production environments solving real problems
(how I wish it was always as simple as Don's examples), I might
unfortunately be mistaken.
I think the preamble where Don yet again prattles on about "Silver Bullets"
is an attempt to save some face following criticisms he received after his
"Silver Bullet: optimizer_index_cost_adj" article. However, considering he's
subsequently re-written the whole thing and retitled it with the "Silver
Bullet" title removed, perhaps even Don might agree that such criticisms
were justified, else why all the changes ? Maybe the "ivory-tower academics"
had a point ... Of course there are silver bullets, here's a big list of
them is still his stubborn claim though. What he fails to realise is that
they're not silver bullets at all but specific solutions to specific
problems which maybe of benefit in specific situations. If for example the
optimizer_index_cost_adj parameter is a silver bullet, why not have used
this silver bullet in all the FTS problems ?
This is where I kinda agree with Jonathan. For a DB article, it actually has
it's good moments in that at least he attempts to narrow down and describe a
specific problem, he attempts to justify an appropriate solution and he
attempts to see the impact and the positive effects of applied solution. He
may have a dislike for those that practice a scientific approach to database
administration but at least he's making an attempt to follow some of the
basic steps.
Where the article falls down for me is the numerous errors that again
proliferate throughout. Time and again DB lets himself down with error after
error after error in an article. It just tarnishes the whole thing and dents
the credibility of the author. The errors also makes the claim that all the
stories are true somewhat questionable in that much of the "evidence" that
supports the "stories" is inconsistent or plain wrong (eg. using alter
system commands to change the optimizer parameters, interestingly now fixed,
the faked statspack reports as it's a little difficult to believe that two
separate databases had issues with identical schemas and table definitions
and even had an identical number of FTS for one of the tables, solving a
different problem to that listed in the top 5 wait events, the faked top 5
wait events as the wait times and % wait times are inconsistent in all the
top 5 wait reports listed, putting tables in the buffer pool that aren't
listed in the report, etc, etc). If the evidence is made up, what does that
say about the rest of the article ...
Finally, the Oracle Support comment. Does Oracle Support really recommend
Don and other external consultants when the customers gets confused with OS
instructions. Hopefully we'll find out as I posted a question on metalink
out of curiosity.
In summary, the article has it's moments and provides some useful pointers
when dealing with simplistic tuning problems. But the ever present errors
and the silly preamble and tone again tarnishes the whole thing.
Richard
----- Original Message -----
From: David Wagoner
To: ORACLE-L (E-mail)
Sent: Friday, March 05, 2004 8:29 AM
Subject: Oracle 911 Article
You guys are going to enjoy this :-).
I just read a very interesting article that managed to insult some of my
favorite Oracle authors and the entire state of Alabama! Now I've heard Mr.
Burleson speak before, and I must admit that I thought he was a dynamic and
entertaining speaker. But do you really think that Oracle Support has ever
told anyone to call HIM to solve their performance problems (his article
makes this claim)? Does anyone else find the tone of this article
unbefitting of an Oracle professional with such extensive credentials? I
honestly can't believe that DBAzine.com published this, especially
considering that Oracle is listed as one of their sponsors.
http://www.dbazine.com/burleson20.shtml
Best regards,
David B. Wagoner
Database Administrator
----------------------------------------------------------------
Please see the official ORACLE-L FAQ: http://www.orafaq.com
----------------------------------------------------------------
To unsubscribe send email to: oracle-l-request@xxxxxxxxxxxxx
put 'unsubscribe' in the subject line.
--
Archives are at //www.freelists.org/archives/oracle-l/
FAQ is at //www.freelists.org/help/fom-serve/cache/1.html
-----------------------------------------------------------------
Other related posts:
