Why artificial test cases don't work
A reader, September 08, 2005 - 8:06 pm UTC
Because the database makes completely different decisions when it sees "real" work going on. If it detects an artificial test case, it will do as it is documented to do, but when you start throwing transactions and queries at it, it turns on fast=true and does something *completely* different.
The unfortunateness is that, much like arguing with religious extremists, no matter how much evidence you show, they will still cite their beliefs as fact.
September 08, 2005 - 8:21 pm UTC
what is artificial about showing how these parameters actually work?
These test cases are not "artificial", they in fact demonstrate that the assertions put forth in the 1st, 2nd, and 3rd edition (albeit less of the assertions over time, since he is incorporting the feedback over time) are not true.
If they want to explain undocumented internals, it is somewhat relevant to get it right.
Note: I did not read source code to figure out the above, didn't need to. I did not ask the developers, didn't need to. Jonathan sure didn't.
There is no such thing as an artificial test case, there are test cases - period. They demonstrate how the thing *works*. Sort of like I did in this chapter:
</code>
http://oracle.apress.com//betabooks/betabook.html?betabook_id=30&APRESSESSID=c5e74fa001f64357db462f5e677af1b1 <code>
scaling up from 1 to 500 users showing how the automatic pga memory management kicks in - reduces the memory allocated to sorts, tries to stay in bounds. That was a "staged" test but demonstrates how it works.
test cases are excellent at showing that "X is not true".
test cases are excellent at showing how things work.
David Aldridge, September 08, 2005 - 10:41 pm UTC
"fast=true"?
I think you must mean the well-known "_fast=true" undocumented parameter, closely related to "_crash=never", "_full_table_scans=evil", and "_detect_artificial_test_cases=true"
Seriously though, if the original article had made claims about PGA sizing and stated that it was true in the context only of a "real-world system" (and also defined what conditions must be met in order for a scenario to be considered a real world system) then you'd have a testable case -- without a testable case all you have is unsubstantiatable rumour and myth. When someone produces a script that explicitly proves the article's claims to be incorrect (or maybe more kindly for there to be some unspecified number of special conditions in order for it to be true) then what is the value of that article? The reader has no idea whether the statements made were true or false, or under what conditions they might be true.
Furthermore when the article is subsequently modified and in the process becomes aligned more closely with corrections to it published elsewhere, then is that not an acceptance that the criticism had a foundation in truth regardless of any "artificiality" in the test case? Maybe not, but I don't know any other way to interpret it.
Consider re-reading above post
Warren, September 08, 2005 - 11:04 pm UTC
Tom,
I think if you re-read the post above, I think you will agree that it was written tongue-in-cheek. I like the fast=true comment (quoting you)!
September 09, 2005 - 6:56 am UTC
;)
ahh, I see now....
Thoughts about test cases.....
Mark J Bobak, September 08, 2005 - 11:50 pm UTC
Well, there are good test cases and bad test cases. If you combine the fact that you don't know everything about how Oracle works (nobody does) with a propensity to see what you want to see, it's quite easy to be misled by a bad or invalid test case. The real talent lies in engineering a solid, repeatable, robust test case that demonstrates (or disproves) what was set out to be demonstrated, without unknown or ill-considered side effects clouding the issues. Folks like Tom and Jonathan have demonstrated time and time again, their excellence in producing such test cases.
September 09, 2005 - 7:00 am UTC
and just to add one thing (or two)
to make them understandable (concise)
to present them in the first place (less errors that way, many times when I open my mouth without having worked it out on paper first - well, I wouldn't have been right)
to make them small (concise) to demonstrate your point
to publish them in the first place (to permit others to see them and say "yeahbut")
Tongue in cheek
A reader, September 09, 2005 - 12:09 am UTC
Yes, the comment was very tongue-in-cheek. I was equating the reference to "real world situation" with a religious doctrine such as the bible. They argue that Oracle will change its behavior the more it does stuff. As far as I know, the only time Oracle changes behavior based on load is when you've set some kind of parameter that tells it to.
The Pga
Yoav, September 09, 2005 - 12:26 am UTC
Hi Tom.
As Allways, very good and detailed answer.
Waiting to your new book ...
Thanks Again
Beating a dead horse
Guest, September 09, 2005 - 6:18 am UTC
It feels so sad to see so many people going into all this trouble to try and get some article corrected,"When the author Cannot correct it!". Imagine if a person, or a group of persons charging $$$ go to a client side turn some parameter on, write in their report to the client justifying $$$ by turning on some paramters based on 'years of experience and real world knowledge', do you expect that person to admit it in a open forum that their change was wrong. Though probably internally they also realise that it was not correct they have to justify their $$$ and hence will have to keep continuing to defend them. So sad.
I was absolutely stunned.....
Bill S., September 09, 2005 - 8:38 am UTC
to see a 230 page "test case" supporting the outrageous claims made in the original article. That kind of flies in the face of the "we don't have to prove anything" philosophy most recently adopted by Don. I also find it interesting that, though he *AHEM* enhanced the article, the references to Jonathan at the end are now gone and no mention is made of who actually *AHEM* convinced Don to enhance the article.
Jonathan's critique of the article was thorough (as usual), well-documented (as usual), and repeatable by anyone who wanted to try (as usual). AND it makes sense and explains quite nicely why you shouldn't play with undocumented parameters unless Oracle requests it.
Everybody here knows this already, but you guys rock. The more that article gets *AHEM* enhanced, the more right you guys look and the more wrong that original article looks.
September 09, 2005 - 8:55 am UTC
The 230 page test case support rewritten assertions not the original, and even then only partially (Jonathan fills in some gaps).
Given the 1st edition of the article said:
<quote>
The Oracle cost-based optimizer will determine whether a hash join would be beneficial over a nested-loop join, so making more PGA available for hash joins will not have any detrimental effect since the optimizer will only invoke a super-sized hash join if it is better than a nested-loop join. In a system like the example in Figure 1, the following settings would increase the default sizes for large sorts and hash joins.
* pga_aggregate_target = 1000m
* _pga_max_size = 1000m
* _smm_px_max_size = 333m
With these hidden parameters set we see a 5x large size increase for parallel queries and sorts
</quote>
not true for either serial nor parallel.....
and given the 2nd edition of the article enhanced that to:
<quote>
The Oracle cost-based optimizer will determine whether a hash join would be beneficial over a nested-loop join, so making more PGA available for hash joins will not have any detrimental effect since the optimizer will only invoke a super-sized hash join if it is better than a nested-loop join. In a system like the example in Figure 1, the following settings would increase the default sizes for large sorts and hash joins.
* pga_aggregate_target = 4g
* _pga_max_size = 300m
* _smm_px_max_size = 333m
With these hidden parameters set we see a 4x large size increase for parallel queries and sorts:
</quote>
again, not so
but the the 3rd edition of the article has
<quote>
The Oracle cost-based optimizer will determine whether a hash join would be beneficial over a nested-loop join, so making more PGA available for hash joins will not have any detrimental effect since the optimizer will only invoke a super-sized hash join if it is better than a nested-loop join. In a system like the example above, the following settings would increase the default sizes for large sorts and hash joins while limiting those for parallel sorts.
* pga_aggregate_target = 4g
* _pga_max_size = 400m
* _smm_px_max_size = 333m
With these hidden parameters set we see significant size increase for serial sorts and a throttling effect for parallel queries and sorts. To see a reproducible, artificial test case demonstrating sort throttling, Mike Ault has prepared a 230 page artificial test case: Validation of Sort Sizes in a Linux Oracle10g Database. However, bear in mind that it only valid for a specific release of Oracle10g, on a specific hardware and OS environment, and not using any optional features such as the MTS.
</quote>
Quite a change isn't it? Gets hard to keep the context straight. We went from a goal of "5x increase" (not achieved), to "4x increase" (likewise), to eventually saying "well, what we really meant was for serial things we'd like to super size and for parallel things downsize".
If only the test case to show the point had been prepared prior to writing about it.....
(you have to love the use of the word "artificial".... Wonder what an "all natural" would be)
I've removed a comment
Thomas Kyte, September 09, 2005 - 9:01 am UTC
Claiming to be from someone who has repeatedly complained that people impersonate them here.
I will continue doing so, so as to not incur further complaints from that individual.
They used an IP address I could not confirm.
And an email address that was not the one they've used in the past.
September 09, 2005 - 9:09 am UTC
However, to whomever made the comment -- you said in it that is was stated, clearly, that the undocumented parameter will require a restart for it to kick in.
I never once in the series of articles saw any mention of a restart being necessary. Maybe I missed it, but I just re-read version 3 and don't see it. If it is included in the 230 page artificial test case, so what. People read the article, 230 pages is sort of big. I don't really see it clearly in there either though.
(and sort of strange that it would be "required" wouldn't it? I mean wouldn't that indicate to you that something else is afoot here - why should a restart be necessary - it isn't by the way, as demonstrated).
Where is freedom of Press?
Janet Burleson, September 09, 2005 - 9:38 am UTC
Freedom of the press is currently being challenged by OracleÂ’s most highly publicized personality. Â Mr. Tom Kyte, The Ambassador of Oracle, has repeatedly demanded that an Oracle tip be removed from the independent portal, SearchOracle.
Tom Kyte had previously published about his control over tips within Oracle Magazine and OTN, and it appears that he is trying to extend his sphere of control to include independent publishers of Oracle related material also.
Mr. KyteÂ’s unwarranted demands for removal of this tip is a serious threat to the freedom of all independent Oracle publishers, especially when tips, articles and books are unfairly condemned by an Oracle-sponsored personality.
Despite SearchOracle's refusal to remove the material in question, Mr. Kyte has published that he does not accept their right of free press and that he will continue to demand the removal, stating
"I am like the energizer bunny, I keep on going and going and going and going, until it is gone and gone and gone.", and a week later Kyte published another call for action: "I just wonder why the article still exists?".
Other independent publishers are also feeling the heat. Mr. Kyte’s blog contains specific suggestions to give all Rampant TechPress books a "one-star" rating and to contact book distributors saying that all Rampant books are inaccurate.  This appears to be a blatant attempt to interfere with fair competition in the publishing marketplace.
It is unsettling that someone from Oracle Corporation has arranged to refuse to stock any Rampant TechPress books in the bookstore at Oracle OpenWorld 2005, even though Rampant TechPress is the fastest-growing Oracle book publisher in the world!
Mike Ault, one of the best selling Oracle book authors and a personal friend, notes his concerns about alleged abuse of power by Mr. Kyte:
"I am especially upset at the labeling of Tom Kyte as the “Ambassador of Oracle”, especially when I receive rude unsolicited e-mails from Tom Kyte claiming that I’m a “Bad DBA”, and condemning me for publishing an article about recovering a lost table column by updating the Oracle data dictionary."
"I have also seen Tom Kyte repeatedly violate the Oracle Code of Conduct and openly insult and condemn his fellow Oracle professionals. He has also allowed others to impersonate me (and others) on his ASKTOM web site and he permits people to make false offensive remarks and publish gratuitous insults."
It is grossly unfair for Mr. Kyte to abuse his Oracle-sponsored fame, power and influence to punish innocent third parties because of a personal dispute.  It's appalling that an Oracle employee would harass an independent Oracle publisher.
Surely these attacks are not sanctioned by Oracle Corporation?
Harasshment
Harry Conway, September 09, 2005 - 9:45 am UTC
Mr. Kyte,
My name is Harry Conway and IÂ’m an author with Rampant TechPress. I see that you have removed some things from your blog, but itÂ’s clear to me that you donÂ’t seem to have much respect for SearchOracleÂ’s freedom of the press. Your blog said that everything in BurlesonÂ’s tip could be proven wrong, and that itÂ’s a fact, not your opinion:
</code>
http://tkyte.blogspot.com/2005/08/getting-credible-information.html
“(they offered to post my findings but I really did not want a public debate, there was nothing to debate here, this was not an opinion piece).”
“the article was just wrong, wrong in its advice and wrong it its interpretation of how PGA memory management works.”
“Everything in that article could have been proven (proven wrong in this case). The assertions put forth could have been tested.”
I’m pretty sure, a lot of things written can be “proven” wrong if you try hard enough and use obscure features and options, even your stuff on ASKTOM. Jonathan Lewis published:
http://www.jlcomp.demon.co.uk/snark.html#Addendum <code>
"I do agree, however, that some of the observations in the original article were “substantially correct” ".
Tom, what gives you the right to demanding that this tip be removed? There is still freedom of the press, and it sure looks like you are going to keep bothering SearchOracle until you get your way. Why wonÂ’t you respect SearchOracleÂ’s right to refuse to remove the tip?
“it is so hard to get wrong stuff pulled from a technical web site.”
“You want to be insulting, you better be 100% correct.”
“I cannot accept it, I don't think I cannot (we cannot) help change it. Maybe I'm hinting - that page has a link to send comments to the editor.... “.
"I am like the engergizer bunny, I keep on going and going and going and going, until it is gone and gone and gone."
To me, it looks to me like you were using your Oracle-anointed fame to harass a competing book author. Are you trying to eliminate competition?
“Good to see people like you are no longer ignoring people like Don Burleson who most of the time post half truths and totally wrong "technical Information". You are doing a great service in pointing out such issues and I know this is not the first time you are doing it.”
“Thank you Tom and Jonathan Lewis for exposing these dreadful, dangerous articles for what they are.”
This sure looks like harassment to me, especially since the original article looked fine to me. Since you said that you did not want a public debate, then why are you publishing this?
“I came up with examples and explanations that paragraph by paragraph showed this and sent it off to the website, asking them to please retract the article. Unfortunately, they did not (they offered to post my findings but I really did not want a public debate, there was nothing to debate here, this was not an opinion piece).”
What is really irritating is that the rigged “proof” offered by Jonathan Lewis that uses shared servers, a fact hidden way-down into the text. C’mon, who uses the MTS on a PC? I’ve been a DBA for over 10 years and hardly anybody uses MTS anymore. You even say that, right here on ASKTOM:
"Unless you have a real reason to use MTS -- don't."
"a shared server connection is by design "slower" than a dedicated server (more stuff goes on, more complex) it is most likely only getting in the way."
What really concerns me is that IÂ’m being damaged because of your fight with Burleson. Your blog published specific recommendations for trashing my upcoming book:
“Contact the distributors of your books (Amazon, Bookpool, Barnes and Noble, Borders) explaining that his books are inaccurate. “
“A lot of one-stars for everything his "Rampant" publishes.”
For the record Mr. Kyte, I don’t appreciate this and I see that your readers are already starting to post “one star” reviews of Rampant books that they have never read, all based on your nonsense about “any single test case proves something wrong”. I know a smear job when I see one.
This week I was informed that somebody from Oracle Corporation ordered Barnes & Noble not to stock any Rampant TechPress books at the Oracle OpenWorld bookstore.
Was that person you, by chance?
Does Oracle Corporation approve of this? I do not, and IÂ’ve already sent a complaint about you to Oracle. I should not be punished because of your continuous unfounded attacks against Burleson.
September 09, 2005 - 10:39 am UTC
Harry --
I removed nothing from my blog. I removed a comment
on this site from a person claiming to be someone, and that someone has repeatedly complained that people impersonate him here (you can get a clue from Janets posting if you like).
I have demanded nothing, if you feel otherwise, please provide a reference to it in writing. I have suggested, I have said to searchoracle "ultimately it is up to you" (making it clear -- their call, I gave them facts to do with what they pleased)
I fully respect Tim DiChiara's right to publish what he chooses (why do you not do the same for me?). I made it clear - it is totally up to them.
Should we perhaps lay version 1 of this article (the cause of all of this) against version 3 of the article and see the *fundemental enhancements* that have been made? It is clear that the author incorporated much of what he read (his email to me said as much) from me and Jonathan. Version 1 was exactly as described.
Compeition is good.
Bad information is - well, bad information.
I eliminate bad information when ever I can. Do you think that is a bad thing?
If the original article looked "fine to you", well, that is precisely why it needed to be pointed out. Lots of misconceptions out there Harry. Lots of them.
A rigged proof? No, more of a slow reveal - showing that hey, maybe there is something to this parameter. It is a nice writing technique.
The fact that you believe hardly anyone uses shared server means we can just totally ignore it?
And remember, this is important, remember - this is, was, never has been about shared server. Did you read Jonathan's original article? There was way more than one point in there:
</code>
http://www.jlcomp.demon.co.uk/untested.html
Also, you sort of missed Jonathan's point with the "substantially correct", if you go back and read it -- you missed the point:
http://www.jlcomp.demon.co.uk/snark.html <code>
<quote>
I do agree, however, that some of the observations in the original article were substantially correct I would go so far as to say absolutely correct specifically: there are some undocumented parameters that few people understand, and changing the _pga_max_size, _smm_max_size, and _smm_px_max_size can affect the memory available for workarea operations.
The devil (as the saying goes) is in the detail.</quote>
tongue in cheek, you might have picked up on that style from his "rigged" example.
Was that person you, by chance? That is business stuff that is soooooo far out of my realm of influence. I'm flattered you think I could do that, but no. I don't control who gets into the bookstore at all.
unfounded Audio pronunciation of "unfounded" ( P ) Pronunciation Key (n-foundd)
adj.
1. Not based on fact or sound evidence; groundless. See Synonyms at baseless.
2. Not yet established.
If we must, we can lay version 1 next to version 2 and see if the issues raised about the article were unfounded or not. It would show the enhancements.
I will say this however, if you give me a list of specific ISBN's which have unfairly received a 1-star review since Jeremy Smith posted a comment to that effect on my blog, I'll look into it. That would not be fair to the individual authors, I totally agree.
Where's the meat (1-star for the usefulness of Janet's post)
Andrew, September 09, 2005 - 9:54 am UTC
Oh, Janet, Janet, Janet... or is it Don?
Like many of Mr. Burlson's postings and articles, lots of words, lost of big statements, but no proof. You wrote many words and have lots of quotation marks, but you have not quoted the target of your invective. Show us what Tom said, when he said it, and where he said it. . . that he _demanded_ that an article be removed. You have other quotations withot attribution. Who said them, when, and where?
Just like you offer no proof that twiddling with undocumented parameters will improve performace, there is no proof of what Tom, or anyone else, has said.
Just Wow
Bob B, September 09, 2005 - 10:10 am UTC
Personally, I'm offended by Janet Burleson's comment on here. This is a technical forum and is no place for political type comments.
Tom,
If you could talk to Janet and ask her to move her comment somewhere more appropriate (i.e. your blog), it would be greatly appreciated. It looks like the fate of this article will be the same as the "Predictive Reorganization", if the political type comments continue.
Freedom of the Press
Andrew, September 09, 2005 - 10:20 am UTC
Harry and Janet,
Freedom of the Press has to do, solely, with the right to publish (or speak) as you choose without fear of being squashed by the big hand of government. Freedom of the Press has nothing to do with printing whatever you wish and not criticized for it. You have the right to print whatever you wish about any subject you wish. You may not, however, claim that it is a violation of your first amendment right to criticize your writing – or even to ask you or your publisher to retract it when presented with proof that it is incorrect.
Consider, for a minute, would your articles pass peer review and be published in the journals of the IEEE or ACM – assuming that they would publish articles on such topics.
Now, as for ‘based on years of real world experience. . .”, and such. I have seen people fired over justifying system changes, solutions to problems, and the like with statements not unlike that one. “Trust me, I am an expert and I know what I am doing. . .” does not cut it in the real world of IT – at least not in the shops where I work. The _ONLY_ thing that works there is “I know it will work because I have tested it, and here are my test cases and results to prove it.”
So far as what I have seen in these discussions, all you are being asked for is proof that only the predicted effect happened, and that suggested change, and only that change, caused that predicted effect.
David Aldridge, September 09, 2005 - 10:25 am UTC
"Freedom of the press" works both ways, folks.
People have a right to post articles, and others have a right to say that the articles are wrong, poorly phrased, not universally applicable. Wander over to my Oracle Sponge blog and you'll find times when I've changed my own postings to accept criticism ... and I've acknowledged the source of the criticisms also.
Your best defence against a charge of incorrectness would be a proof that you were correct, but having tacitly accepted the criticisms by modifying the article twice (without direct acknowledgement of the source of the corrections) that is no longer an option. Hence, I guess, the new-found dedication to the cause of "an open and free internet" -- a remarkable contrast to the amount of banning and post deletions that have occured on the forums associated with Burleson Consulting in the past.
I recall a page on the BC web site stating that for legal reasons BC consultants were not permitted to post on third-party forums and such-like ... unless I'm remembering that incorrectly, TK's caution in accepting that people who post under the name of a BC consultant seems eminently justifiable.
Bending of the truth
Bob B, September 09, 2005 - 10:26 am UTC
In Janet Burleson's comment above she has bent the truth
(at least once verifiably, probably more). Here is a link to the article she quoted.
</code>
http://www.jlcomp.demon.co.uk/snark.html <code>
Here is Janet's quote of the article:
"I do agree, however, that some of the observations
in the original article were “substantially correct”".
Johnathan continues on saying that the part that was "substantially correct" was that there are undocumented parameters and changing these parameters can affect the memory available for workarea operations.
The following quote is the entire paragraph surrounding the quote made by Janet - proving my point above:
"In this context, I believe the word “enhanced” has been
used with its special meaning of “corrected”. I do
agree, however, that some of the observations in the
original article were “substantially correct” I would go
so far as to say “absolutely correct”– specifically:
there are some undocumented parameters that few people
understand, and changing the _pga_maxÂ_size,
_smm_max_size, and _smm_px_max_size can affect the
memory available for workarea operations. The devil (as
the saying goes) is in the detail."
Correction
Bob B, September 09, 2005 - 10:28 am UTC
Meant to say "Harry Conway" above, not "Janet Burleson". My apologies
I found it! Documentation of FAST=TRUE (tic)
Andrew, September 09, 2005 - 10:38 am UTC
From the (not)Oracle Reference guide to undocumented features and parameters:
_FAST
Parameter type : Boolean
Default value : Derived
Parameter class: Transactional Dynamic
Range of values: true | false
_FAST enables Oracle to determine whether (or not) to execute the query quickly or slowly. _FAST replaces, and its value is opposite to that of, the deprecated parameter _SLOW. _FAST is derived from the explicit settings of certain combinations of other initialization parameters that govern the optimizer, memory, and IO as well as the structure of the query, tables involved, and any indexes present – or not present that would be used if they were present. A significant part of what the database uses to set this parameter is the presence, or absence of, complete and correct statistics on the tables and indexes involved in the query/transaction. To some extent the dynamic setting of this parameter is governed by the concurrent load on the database, but no so much as you might think.
Even though the _SLOW parameter is no longer used, if you built you system so as to cause _SLOW = true, then (if you leave things as they are) the system will never set _FAST=true, no matter how many CPUs or how much memory you add.
Although not explicitly set, the _FAST parameter is completely under the control of the application developer and database administrator. You cannot see the _FAST parameter in any dictionary table or view. The only way you can know if it is set to TRUE or FALSE is by measuring the performance of your application and the end-user experience.
See Also:
- Oracle Database Concepts for a discussion of how Oracle works.
- Oracle Database AdministratorÂ’s Guide for more information on determining how to set modifiable parameters.
- Oracle Education Catalog for more information on classes available on writing well performing applications, query tuning, and database tuning.
A reader, September 09, 2005 - 11:32 am UTC
"Andrew from Michigan, USA" that was hillarious. :-)))
Rahul, September 09, 2005 - 11:39 am UTC
I agree with Andrew (from Michigan, USA ) and everybody we need proof. People need more than the same "real world experience trust me.. da da da.. worked on large system .. da da da.. do do de de de"..
Confused, September 09, 2005 - 11:41 am UTC
I just want someone to please explain to me how criticizing an article that is shown to be demonstrably wrong (the original article, and the second iteration of it) is an attack on freedom of the press? I thought that you were supposed to be able to refute statements that could be shown to be wrong by providing the evidence that proves the statement wrong. Seems to me Jonathan and Tom were able to do this without difficulty. So where is the attack on freedom of the press? By asking to have a provably incorrect article retracted? I'm puzzled.
Dear Confused:
Andrew, September 09, 2005 - 11:51 am UTC
The first ammendment to the US constitution used to read
Amendment I (1791)
Congress shall make no law respecting an establishment of religion, or prohibiting the free exercise thereof; or abridging the freedom of speech, or of the press; or the right of the people peaceably to assemble, and to petition the government for a redress of grievances.
It now reds (or at least some would like it to read)
Amendment I (Rev 2005)
Congress shall make no law respecting an establishment of religion, or prohibiting the free exercise thereof; or abridging the freedom of speech, or of the press (unless I wrote (said) it and do not like criticism, in which case I get to abridge your rights otherwise herein guaranteed) ; or the right of the people peaceably to assemble, and to petition the government for a redress of grievances.
Things look a little different in context:
Still Confused, September 09, 2005 - 12:18 pm UTC
BTW, Janet, the full energizer bunny quote was as follows and can still be found in Tom's blog archive:
[quote]
When I see wrong material, that many might read and just believe because "a name" said it -- or even just because a respected website printed it -- I am like the engergizer bunny, I keep on going and going and going and going, until it is gone and gone and gone.[/quote]
Note the emphasis on WRONG MATERIAL. Not just any material by any one specific person. It isn't a "personal attack", never was.
You might consider not taking statements out of context to try and prop up your position.
First amendment
Bob B, September 09, 2005 - 12:27 pm UTC
Hypothetical here:
If Tom went to court to get the article removed, that could be a violation of 1st amendment rights. If Tom goes to the publisher and says "Please take that off the sight" (or something to that effect) and they do, that is not an infringement of the first amendment. Tom is not the government and is therefore protected by the 1st amendment, not limited by it.
Hidden Parameters
Greg, September 09, 2005 - 12:40 pm UTC
It IS interesting that the original article is recommending that a user play around with hidden parameters. Here is a quote from a book(let) that I was given:
"... A hidden parameter is one that is not listed in the Oracle documentation, but that still exists for use by Oracle Support Services."*
That seems to be a "Don't do that" warning.....
* From: Unix for Oracle DBAs, Pocket Reference
Donald K. Burleson
Back to the devil
Scott Swank, September 09, 2005 - 12:57 pm UTC
... by which I mean back to the details. My problem with Don Burleson's work is that he does not pay enough attention to the details. If he were to say things like:
"In version 9i & 9iR2 I performed the following with the following relevant setting. These are the results I saw. This leads me to believe X."
Then I do not believe much of anyone would have an issue with him or his work. People would undoubtedly point out from time to time that there are additional relevant settings or that different behavior is seen with a particular patch, etc.
The problem is that he does not do this. He makes sweeping claims, some of which clearly indicate that he has NOT done the relevant work. And for this reason I, personally, have come to distrust his statements.
He simply needs to put in the due diligent work and state no more nor less than he has seen. He needs to needs to make clear what he's assuming and what he knows for a fact to be true. This is not too much to ask. Were he to do this I imagine he would see his legion of so-called enemies disappear.
Cheers,
Scott
Jeff Hunter, September 09, 2005 - 2:58 pm UTC
"Tom is not the government", only "The Ambassador of Oracle".
September 09, 2005 - 3:12 pm UTC
sigh, I did not make that up trust me.
Nope, I believe I'll stand by my original statement
Bill S., September 09, 2005 - 2:59 pm UTC
Tom,
I originally said I was stunned
[quote]to see a 230 page "test case" supporting the outrageous claims made in the
original article.[/quote]
To which you replied:
[quote]The 230 page test case support rewritten assertions not the original, and
even then only partially (Jonathan fills in some gaps). [/quote]
But, Don's note in his own article says (direct copy and paste just a few moments ago):
[quote Don Burleson's article]
I have enhanced this tip again, including more clarifications and a 230 page artificial test case to show that my original tip observations were substantially correct.[/quote]
He deliberately said "to show that my ORIGINAL tip observations...." So he claims that the test was based on the ORIGINAL article as written, not the revised one.
I'm still stunned.
September 09, 2005 - 3:16 pm UTC
Bill -
I was just pointing out that the supplied test case shows the original "facts" presented to be false, not true. I know what he said and Jonathan did a good job with the "devils in the details"
That is why I added the copies of the 1st, 2nd and 3rd editions (the test case was about those points) when following up your comment. The test case supports part of the 3rd release but shows the first two "more incorrect"
Now, since all three say very different things...
Got it now. Thanks for the follow-up!
Bill S., September 09, 2005 - 3:22 pm UTC
Tom,
Got it, took a few read-throughs for it to sink in. Thanks for correcting me. I appreciate it. :-D
Like you said,
"Gets hard to keep the context straight."
Bluster
Jim, September 09, 2005 - 3:26 pm UTC
Assuming Janet wrote the above, and she may not have, it is typical legalese bluster. The spin is so high on this one that we could use it to power a small country. Either she is not very intelligent and doesn't have the basic understanding of what freedom of the press is or has such a large ego that legalese bluster is the only answer to her. I believe that it is the latter. (no evidence, this is just my opinion)
It really is sad. This behavior is typical of bad polititians. (blame others, don't correct your behavior) It takes a humbleness and a high degree of professionalism to admit you have made a mistake and correct it. Surprisingly people react to such professionalism with sympathy and respect. We all make mistakes.
There are 4 types of people in business doers, checkers, fackers, and takers. Doers are those people in the trenches that do the actual work. Checkers are those that guide the enterprise and thus guide the doers. Fakers are those that look like doers, but are a drag on the enterprise. Takers are similar to fakers, but try to take credit for other's work.
When people find a faker and a taker the backlash is great.
Don and Janet, learn to be professional. Correct your mistakes in a professional manner, give credit to those that have pointed out an error and fix the error. This legalease bluster stuff is not doing you any good; it isn't professional and people see right through it.
September 09, 2005 - 3:41 pm UTC
I can only verify that the IP address used was one that has been used to send email to me in the past from Burleson Consulting and it is registered to them.
I would have removed it with a note otherwise, like I did the prior entry.
There was a news article on their website as well that contained text very similar to this. (contained, past tense, it was there and then gone but I've compared the two and they came from the same place)
I did not feel the need to dignify it with a response.
A reader, September 09, 2005 - 4:04 pm UTC
Its amazing how the Burleson clan go around bullying everyone.. I can't believe the repeated unprofessional behavior and insensitivity to legitimate criticism. I just wonder what kind of nutheads do hire them for doing work...
One line that stood out in Janet Burleson's post was about "feeling the heat". Surely they are feeling the heat and unless they learn to behave like professionals they are going to feel it even more.
A reader, September 09, 2005 - 4:26 pm UTC
<JB>
Freedom of the press is currently being challenged by OracleÂ’s most highly
publicized personality. Mr. Tom Kyte, The Ambassador of
</JB>
When did Larry give up his designation as Oracle's most highly publicized personality? ;)
<JB>
Mr. KyteÂ’s unwarranted demands for removal of this tip is a serious threat to
the freedom of all independent Oracle publishers,
</JB>
A wrong tip can pierce your skin and damage you. It's better to get rid of a wrong tip.
<JB>
It is unsettling that someone from Oracle Corporation has arranged to refuse to
stock any Rampant TechPress books in the bookstore at Oracle OpenWorld 2005,
even though Rampant TechPress is the fastest-growing Oracle book publisher in
the world!
</JB>
I hope they don't make the hotel cancel my reservation for OpenWorld!
September 09, 2005 - 4:39 pm UTC
That last statement is an outright falsehood.
I am very flattered they think I have that degree of control - but I do not at all.
And now they know in no uncertain terms it is false - so I hope to not see it stated again.
A reader, September 09, 2005 - 7:33 pm UTC
"fastest-growing Oracle book publisher in the world!"
Thats so typical DB.. Self proclaimed worlds best Oracle DBA and now for publisher turn..LOL...
Spreading wrong knowledge???
A reader, September 09, 2005 - 8:25 pm UTC
For Janet, Don, Mike etc...
Please (to be nice - for the respect of Tom) don't spread wrong knowledge, as if you continue to do so, it could be treated as spreading rumors!
Thanks
Learning Oracle - Don Burleson style
Enayet, September 10, 2005 - 3:42 am UTC
Dear Don,
My first book about Oracle from Oracle Press by Don Burleson. I salute thee!
It's been a while since I bought my first Oracle DBA Handbook and in my dismay, I have learned lots of 'bad knowledge'. I note few examples here:
1. During hot backup Oracle do not write to datafile, generating excessive redo logs!! -- wrong :((
2. ora-1555 - snapshoot too old error happen due to smaller rollback segment!! -- wrong (partially) :((
3. super sizing PGA with _parameter -- wrong :((
The list can go on.....whom should I blame for these 'bad knowledge' inflicted on me? It's people like Don et el.
I know you have been tempted to toy with hidden parameter, feeling the wrath of Steve Adams, J Lewis, and most notably Oak Table groupies. I am sure you, Don, has a desire to proclaim the BEST Oracle dude on this planet. The desire to cling onto your ego is good, BUT foray of evil intention is BAD!
I also know that you have written numerous books about Oracle (25 or more?). So there might be a 'weired' connection lies here. Seeing your empire of Authority is in peril might hurt your ego!! But, hey, it's a very hard truth!
Saying a wrong thing is just bad, spreading it - is even worse.
Don, please stop these bad vibe - PERIOD.
'cmon now
Jeff Hunter, September 10, 2005 - 6:51 am UTC
[quote]
It is unsettling that someone from Oracle Corporation has arranged to refuse to
stock any Rampant TechPress books in the bookstore at Oracle OpenWorld 2005,
even though Rampant TechPress is the fastest-growing Oracle book publisher in
the world! [/quote]
If you've dealt with Oracle for any time, you'd know this was false. IF they have excluded Rampant TechPress, you can bet it comes down to dollars and cents instead of innuendo and finger-pointing. If Oracle thought they could make a buck on it, you would have a big Rampant TechPress booth.
Alex, September 10, 2005 - 1:17 pm UTC
Tom,
I honestly feel bad for you, that you have to deal with this. I must be like fighting with your children. I remember I did it when I was a kid, when I knew I couldn't disprove my parents I'd just make stuff up ("Demanding to have the article removed....").
There's one thing I don't understand through all of this; Why don't they give Jonathan Lewis the same treatment you get? He slams Don much, much worse than you do is his article. despite that, they quote his article as an attempt to support themselves?? (As you pointed out, I think they missed the boat on that one). The same went for the previous debate on askTom about reorganizing I think? It's like they're just pretending John's on their side or he's off limits for some reason. I don't get it.
Possible answers for Alex
Maybe Not So Confused, September 10, 2005 - 2:37 pm UTC
Alex,
Jonathan doesn't get the same "treatment" (and this is just supposition on my part of course) because he is in the UK and doesn't have the same exposure in this country that Tom has, and Tom is also a better target because he works for Oracle. So Tom, being more visible and a more "public" authority here, gets more grief. Jonathan is obviously brilliant, obviously knows Oracle, and obviously knows how to show people what is wrong with their own "understanding". So is Tom; but Tom is here, and Jonathan is there so Tom is easier and better "press".
Just my take on it - only worth $.02 (US, not Canadian ;-D)
TV debates
Alberto Dell'Era, September 10, 2005 - 2:45 pm UTC
> must be like fighting with your children.
Actually, it reminds me of some political debates on TV - when a politician is unable to win the debate using logic and reasoning, a well-known technique to try to get some points home is to shout and try to turn the debate into a verbal fight ... in order to move the attention of the audience away from the other's winning arguments.
I've trained my mind to switch off to standby during the fighting parts (and a friendly timeout automatically switches off the politician entirely after a while); never thought that this feature of my brain would have proven useful in technology-related discussions (nice example of cross-domain transfer).
Relevant
Conno, September 11, 2005 - 9:04 pm UTC
undocumented parametes,
sns, September 12, 2005 - 11:20 am UTC
I haven't heard about _pga_max_size and _smm_max_size so far.
Anyway, on my 9.2.0.6 database I got the following:
SQL> select value/1024/1024
2 from v$parameter
3 where name = '_pga_max_size';
no rows selected
Elapsed: 00:00:00.01
SQL> select value/1024
2 from v$parameter
3 where name = '_smm_max_size';
no rows selected
What does this mean?
Thanks,
September 12, 2005 - 2:16 pm UTC
in order to see undocumented parameters, you must use an undocumented table ;)
select x.ksppinm name, y.ksppstvl value
from x$ksppi x , x$ksppcv y
where x.indx = y.indx
/
A reader, September 12, 2005 - 11:53 am UTC
Maybe hidden parameters are HIDDEN parameters ;-)
PS: to find something hidden you may have to look at some hidden "places".
Burleson is right out of Harry Potter...
Ryan, September 12, 2005 - 1:45 pm UTC
Don and Janet Burleson remind me of the Dark Arts teacher from the second Harry Potter book or movie. The guy talked about himself continually to make people think he was an expert to sell his books and make money. He was just a big phony. However, he was smart enough to know he was a phony.
Looks like Janet Burleson is doing the same thing with 'Guide Horses'. Howard Rogers has some excerpts on his blog about her lack of knowledge while acting like an expert from someone completely outside of Oracle.
</code>
http://www.dizwell.com/blogindex.html <code>
Has anyone else noticed that in order for Don Burleson to have published as many books as he claims to have published he must be including each new 'version' of his book. Well I read parts of his Oracle Unix 9i book. The cover says that it has information on 9i features. It never mentions 9i, it still has information directly pointing to version 7. It doesn't look like he even looked at it, but is taking credit for a new book. This book has an entire chapter copy and pasted from his statspack book.
The Oracle Job Interview book used to come on a CD with Mike Ault's Admin and Management book. They just took that and put it in a book. If I remember correctly many of the questions still refer to version 7, but its sold as a new book.
Anyone else notice the size of the print in Rampant press also? I have seen $60 price tags for books with the same material from other presses that sell for $30, but have larger type.
The Utilities book has NOTHING in it that is not in the Oracle Utilities guide. Go flip through these books yourself. You can grab them at most bookstores. I have not read all the Rampant press books so this is not an attack on everyone who publishes for them. However, I won't waste my time reading those books either.
Don had on his blog that he and his family certified as a diver. How long until we see 'Burleson Diving: The World's Formost Diving Experts'.
I typically just ignore the guy.
September 12, 2005 - 2:31 pm UTC
Well, I will not/would not condemn the entire series of books here. I know many of the authors either "electronically" or personally (Arup Nanda for example, really really good stuff).
So, I would not condemn the entire suite of books in that fashion - it isn't fair to the authors in general. Books take hard work. You get little in the way of return financially.
I judge a book not by the cover, but by the author and review teams. A good author, with a good review team - a good book makes.
you are correct
Ryan, September 12, 2005 - 2:56 pm UTC
It's fair not to condemn all the Rampant books. The problem is that so many of them are just rehashes of documentation that I can get for free or totally inaccurate. How do you tell if they are good?
I would rather just avoid paying money for Rampant Books since odds are it's not worth it. Tom, you have an article on your Blog today about Trust. Rampant has little trust. Don Burleson's name on it gives it less trust. So why should I invest my money in it? If someone has it laying around the office I may flip through it, but I won't pay for it.
To be fair, Arup Nanda's material is pretty good and it's clear that he works hard on it. I would be interested in hearing what Arup has to say about Don's work in general. I have noticed that he has not come to his defense.
Also to be fair, Mike Ault's book on Admin and Management is pretty good. However, to be fair in general most books are rehashed of material that is already available either for free or by previous authors. Most articles are rehashes of the documentation. Very little of the literature is worth reading.
September 12, 2005 - 3:20 pm UTC
I trust the author and the review teams - not the publishing firm when I look at a book.
I would trust the general theme of the reviews for or against the book (well, the reviews that come from people not directly associated with the books that is).
One other point
Ryan, September 12, 2005 - 2:58 pm UTC
This is in response to Janet Burleson stating that people are giving one star responses to books they have not read. That is fair. However, Don Burleson and many of his authors give each other glowing reviews on mediocre books just to help each other sell copies. Doesn't make up for the knuckleheads though.
BTW, your preview button doesn't work. Getting an error page.
September 12, 2005 - 3:23 pm UTC
It was Harry that had that comment and I've offered to look into whether they were malicious or not and see what could be done, but no list of books has been forthcoming as yet.
I just tried the preview, it worked for me, what error do you get?
Literacy versus Credibility
Elizabeth, September 12, 2005 - 3:19 pm UTC
The first time I went to Mr. Burleson's site I was amazed at the number of grammatical mistakes I saw. If a professional does not care enough to have his copy proofed (he needs to re-read his second grade lessons on the difference between plural and possessive), how can I give credence to what he says? For example:
BC Oracle performance guru's can quickly verify
every aspect of your Oracle database and provide a
complete certification that your database is fully
optimized.
Why undocumented parameters are hidden.
Brian Byrd, September 12, 2005 - 3:35 pm UTC
If undocumented parameters were generally useful and safe, I would think that Oracle would document them. There must be a reason why they are hidden. In my humble opinion, it is better to start with the default parameters and make changes as system testing indicates. Looking for quick and easy parameters changes that will make SQL run faster is like looking for the foutain of youth. No amount of hardware and changes to database parameters will make up for a poorly designed and or written app.
Review Team
Michael, September 12, 2005 - 3:45 pm UTC
There are a number of "Oracle Experts" who just churn out stuff in such quantity that you KNOW they are just cutting and pasting from Oracle Manuals or have ghost writing monkeys turning out this stuff for them. And there are apparently enough publishers willing to publish this stuff, even without any review team to ferret out the garbage.
I used to work with a former Marine, and his favorite expression was "If you throw enough poop (he didn't say poop) against the wall, some of it's likely to stick". Short and to the point, and it applies to a lot of the books and articles I've seen by so-called Oracle experts.
I think Tom hit on it before, if you have a really good and tough review team, you are much less likely to turn out donkey-dust with your name on it. Good reviewers will smoke you out and bust you if you try to put stuff out that can't be verified. The reviewers should be highly experienced, have the time to properly review the material, and test it. The author should be able to accept any valid criticism offered by them and modify the work accordingly. Otherwise, the book will suffer.
A quick response to Harry Conway
Peter K, September 12, 2005 - 6:35 pm UTC
<QUOTE>This sure looks like harassment to me, especially since the original article
looked fine to me. Since you said that you did not want a public debate, then
why are you publishing this?<UNQUOTE>
Take a look at the original article again. Ask Don for it. The fact that it has gone through a couple of revisions, I would say that Don has corrected or "enhanced" as he calls it, would tell you that the original article is not fine.
The fact of the matter is that Don has published an article with which was incorrect with half-baked assumptions/assertions (see Jonathan Lewis' analysis of said article).
Your stance seems to be that Tom has a personal vendetta and you came on with a bunch of accusations against Tom which are totally unfounded.
Can't answer questions?
Cancunkid, September 12, 2005 - 10:01 pm UTC
My. Kyte,
It just seems odd that on the splash page for this forum you say you don't have time for additional questions but you seem to have plenty of time for all of this tripe.
How about some real answers to some real Oracle questions. I just visited the site:
</code>
http://dba.ipbhost.com/index.php?s=3981ab9ecefb37d4c0867b7171496aca&showforum=2 <code>
They seem to spend time giving reasonably good answers to real questions there, instead of wasting time.
September 13, 2005 - 11:43 am UTC
great idea, go ahead? No one is stopping you.
If you get an answer from David Aldridge or the like you might even get something useful.
Hey kid - patience has it rewards.
Diogenes, September 13, 2005 - 12:20 am UTC
You'll find the more you frequent that forum you mentioned, the more you will come to realize that "good answers" isn't what it's about. If you want "good answers", you need to wait your turn here or try one of the other sites mentioned in the links here.
The reason why they have time for answers over there is that they don't always bother to do the research - yes, some of the folks who post over there are reasonable and are good at what they do - but it is overshadowed by the bad advice that permeates that place. They'll give you answers over there - sure. Tom will give you RIGHT answers. And show you how he got them. Worth the wait, believe me.
A reader, September 13, 2005 - 12:56 am UTC
"Cancunkid from NJ" do you write the same to DB on that forum as "My. Kyte," :-)
Cancunkid - Tom surely can answer questions
Jurij, September 13, 2005 - 7:58 am UTC
Cancunkid, if you look at the top of AskTom main page, you'll find his working statistics. Right now it says:
"In the last 4 weeks, I've taken 105 new questions, read 1,197 followups, and responded to 903 of the followups"
To me it sounds pretty impresive work for one person. I doubt you'll find similar results on that other forum, not to mention the level of quality on both....
Advice just as good
CancunKid, September 13, 2005 - 7:59 am UTC
It is odd you say such a thing. I found many useful answers to my questions there. More than here. As to errors, I see many times here where Mr. Kyte has said he has made an error and had several times his advice, while probably correct, I could not apply it to my database. Rewrite of the application is no option. I do not know if you are currently working or not, but in the place I work we need answers now, not when someone has the time to answer. We also need answers that don't involve thousands of dollars of consulting or many days work rewriting code.
September 13, 2005 - 12:24 pm UTC
They have good silver bullets over there, have at em.
Nobody can make you read one thing or the other - you pick what you want to read, I've no problem with that.
(see the time I spent reading this diatribe - I could have been taking another question ;)
To CancunKid
djb, September 13, 2005 - 8:21 am UTC
You do what you gotta do, kid. If you got yourself into a situation and can't fix it, then you look for help. If you can't afford help, then you get what you pay for.
Do everybody a favor and stop trolling. It does no good here.
Responding to your response
Harry Conway, September 13, 2005 - 9:55 am UTC
Tom Kyte responded: “I removed nothing from my blog.”
Yes. Tom, you did “5 of them - they are gone now.”. You claim that they were spam, but we will never know, will we? You also allowed one comment to be removed by Gabriel Paulovic, without any explanation.
Tom, did you contact Mr. Paulovic and ask him to remove his comment? Because it was “wrong” like you did with the tip? Or maybe you thought that Paulovic’s comment was inappropriate, unprofessional, or not relevant to the discussion? No, that cannot be the reason because you allow people to contribute all sorts of unprofessional contributions to your blog. What did Mr. Paulovic contribute that was worse than this?
- “If only I were the editor, I would publish nothing from Don Burleson. Rest assured that the real Oracle developers here listen to Tom's preaches.”
- “Not sure I can understand why any company would want to do business with Burleson consulting,”
- “You keep saying something totally wrong - people would ask what do we do to make this guy keep this mouth shut. That’s exactly whats happening” .
- “I'm not sure the best way to tackle the independent companies and consulting gigs he does, but I have to imagine that if you talked to the folks at the various conferences you could convince people to wipe him from the slate.””
Tom Kyte Responded: “I have demanded nothing, if you feel otherwise, please provide a reference to it in writing.”
No problem, all of this is published on the Tom Kyte blog. Perhaps I used the wrong word saying that you ‘demanded’ the removal. It looks like a stronger word than “demanded” is in order:
- “I cannot accept it, I don't think I cannot (we cannot) help change it. Maybe I'm hinting - that page has a link to send comments to the editor.... “.
- “When I see wrong material. . . I am like the engergizer bunny, I keep on going and going and going and going, until it is gone and gone and gone.”
- “But I was disappointed the article remains and other people will in fact read it, and maybe believe it.”
- “I just wonder why the article still exists?”.
Tom, at he very least you “demanded” to have the article removed, and I’d bet that SearchOracle would say this as a “demand” or perhaps a stronger word. C’mon, Tom, you encouraged people to repeatedly contact SearchOracle (“Maybe I'm hinting - that page has a link to send comments to the editor....”), all based on your FALSE statement that the tip was “JUST WRONG”.
As I’m learning, it’s sad really that when you say that something is “bad”, “wrong” and “dangerous”, people might listen. Burleson’s original tip WAS NOT completely wrong, and I’d like to see you apologize for publishing this:
- “there was nothing to debate here, this was not an opinion piece”
- “Everything in that article could have been proven (proven wrong in this case). The assertions put forth could have been tested.”
Tom Kyte responded: “If the original article looked "fine to you", well, that is precisely why it needed to be pointed out. Lots of misconceptions out there Harry.“
Tom Kyte, please don’t insinuate that I liked this tip because I’m suffering from some “misconception”. You are the first Oracle employee I’ve ever met who doesn’t treat people with respect, and I find your response to be presumptuous, arrogant and insulting. There was no misconception here and maybe I deserve an apology.
Tom Kyte responded: “I will say this however, if you give me a list of specific ISBN's which have unfairly received a 1-star review since Jeremy Smith posted a comment to that effect on my blog, I'll look into it. “
Tom, you can look them up yourself. There are plenty to choose from and they are easily spotted because the book review dates correspond to your condemnations.
Tom, when you start a blog and condemn someone, tell me that you are surprised when you get dozens or me-tooÂ’s publishing insulting and unprofessional comments. You arenÂ’t fooling everyone, Tom.
September 13, 2005 - 12:41 pm UTC
they were spam harry, face it - look at the discussion, do you see what was immediately following that? Would you like to see them:
1) Your blog could get you recruited - or fired
Terrence Ryan knew Scott McNulty in passing at the University of Pennsylvania's Wharton School, ... and Sun Microsystems Inc.
Nice blog site, really cool!
I have a poker texas
site/blog. It pretty much covers poker texas related stuff.
Come and check it out if you get time.
2) Call it WSU Tech: School at forefront of e-education
PULLMAN + The technology wing of Jeff Records' life is significant. He has computers, both desktop and laptop.
Hi, I just bookmarked you - nice blog! I have no blog yet - lack of time, really... but in stead I have a garden rose site. Rosegardening is fun! Well - feel free to come and check it out if you get time :-)
3) Your blog could get you recruited - or fired
Terrence Ryan knew Scott McNulty in passing at the University of Pennsylvania's Wharton School, ... and Sun Microsystems Inc.
Used Car Loans
4) What a neat blog! Please have a look at this
page too!
5) Thanks for putting up this blog. I run a website for blog related site. It isnt anything special but you may still find something of interest.
Interesting stuff no doubt. If you read the blog harry you would have seen clearly someone said "holy spambots" and that we discussed how to turn on the spambot tracking.
Gabriel had written:
-----------------------------------------------------
Note:223299.1
If PGA_AGGREGATE_TARGET is set in the init.ora, then SORT_AREA_SIZE, HASH_AREA_SIZE, BITMAP_MERGE_AREA_SIZE and CREATE_bitmap_AREA_SIZE are ignored.
It's not just Don who is wrong ...
--------------------------------------------------------------
and why should I "prevent" squat? Why should I prevent someone from removing their comment (I suppose he removed it because others had already pointed out the metalink notes)
You would have to ask Gabriel why he decided to remove his comment, I don't contact people to ask them to retract a comment on my blog - if it was technically inaccurate I would certainly followup with the evidence and reasoning as to why it was - but it would remain.
tell me harry, what is wrong with pointing out that the "tip" has a link on the bottom for to send a letter to the editor with a comment. The "tip" was provably *wrong* -- really realy wrong -- especially at that point in time prior to any "enhancements" (love that - enhancements)
harry, employee of don, do you really want version 1 layed out next to the current version and have all of the contradictions painstakingly pointed out - point by point?
did you notice the article goes from claiming a 5x increase in memory used (in fact no for serial and less for parallel), to a 4x increase in memory, to a
"With these hidden parameters set we see significant size increase for serial
sorts and a throttling effect for parallel queries and sorts."
hmm, interesting - isn't it. especially with a "230 page artificial test case" that "proves the original article was substantially correct"?? No, what the test case shows is the first two drafts certainly got that point wrong, the tip showed how to downsize your PGA.
Harry, if you have specific books - just take a second to point them out to me. There is something like 134 rampant books on amazon.com, I certainly don't have time to look at each one. You must have a list, else you wouldn't have make this claim.
This is not for kids, and you learned it the hardway - come back when you are grown up!!
A reader, September 13, 2005 - 10:01 am UTC
A reader, September 13, 2005 - 10:13 am UTC
Tom, I think you should just ignore these guys - they are never going to get it. It is not worth your time or effort to reply to these stupid remarks...
I think this is going nowhere and Tom's time is too precious to be wasted on such arguements. The best thing to do would be to lock down this thread just like the predictive reorganization thread and avoid any references to Burleson and co. in future.
Discerning readers know who deserves their trust. They dont need to go through this once again to judge who won the debate!
ANNOUNCEMENT: Energizer Bunny has died
Billy, September 13, 2005 - 10:33 am UTC
SAPA.
September 13 2005 at 12:00PM
Subject: Energizer Bunny has died
Immediate Release.
(Cape Town Reporter)
It is reported that the Mr. Energizer Bunny was announced dead on arrival at the Cape Town Hospital. A spokeperson for the hospital said that the death has been contributed to sexual exhaustion.
It was discovered that the batteries of Mr Bunny had been inserted the wrong way. Instead of keeping on going and going, Mr Bunny just kept on coming.
==
PS. Moral of the Story. Make sure you have your batteries the right way in when writing 'expert articles' and making 'expert comment'. Something that some (like Don) seem to forget at times.
So Harry - where's YOUR proof?
Bill S., September 13, 2005 - 11:23 am UTC
Harry,
You said:
[quote]
BurlesonÂ’s original tip WAS NOT
completely wrong, and IÂ’d like to see you apologize for publishing this:[/quote]
Tom published in SEVERAL places why the ORIGINAL article was wrong. Jonathan Lewis published several articles refuting the "original" article. So now you come here and claim the original article was just fine - WHERE'S YOUR PROOF?
Frankly Harry, I'd like to see YOU apologize for your unfounded accusations regarding removed blog entries (in case you didn't know it, people who post blog comments usually have the ability to remove them themselves and occasionally do) and alleged attempts to subvert freedom of the press. But I don't expect to see it anytime soon. I for one find it very interesting that several well-known Oracle experts were able to systematically pick that article apart from the beginning, and yet you claim the "original" article was soooo right and spot-on (but it had to be revised twice based on the findings of these well-know experts).
You also said [quote]C’mon, Tom, you encouraged people to repeatedly contact SearchOracle (“Maybe I'm hinting - that page has a link to send comments to the editor....”), all based on your FALSE statement that the tip was “JUST WRONG”. [/quote]
OOPS - you did it again, claiming that Tom's assessment of that article was FALSE, yet you offer nothing up to substantiate your claim. So who are you kidding? Defend the article all you want, but don't expect people to believe you just because you say so - offer up some evidence if you're right. And don't expect us "working" DBAs to regard the suggestion that it's OK to change undocumented parameters as anything more than recklessness. There is a good reason why the parameters are undocumented, and also there is a good reason why we pay for support. Someone new to Oracle databases might think it's cool to change a hidden parm and potentially reduce their database to an unsupported mess - I wouldn't, not without Oracle Support's insistence.
Tom and Jonathan (and Howard Rogers and everyone else) have a right to challenge incorrect material being published about Oracle. They ALWAYS provide something to support their positions.
harry about tom's removing stuff from his blog
Ryan, September 13, 2005 - 11:38 am UTC
I saw the stuff he removed. They were advertisements to buy things. Has nothing to do with Oracle or anything technical. I saw 3 of them on his site.
I am sure Don gets the same thing and has removed them also...
September 13, 2005 - 1:03 pm UTC
Don doesn't permit commentary on his material.
relevent
A reader, September 13, 2005 - 11:39 am UTC
Responsible Debate
Daniel Fink, September 13, 2005 - 11:43 am UTC
In the interest of full disclosure, I am not an employee of Oracle corporation, nor have I ever been. I have no relationship with Oracle corporation other than as a consumer of their products. I have no connection with Tom Kyte other than in a non-employment professional sense as a fellow member of the Oracle user community. If requested by Tom or another employee of Oracle corporation, I will be glad to remove this post.
Mr. Conway, why did you not also disclose that you are an employee of Burleson Consulting as it clearly states on your resume? This information is different from your previous posting that you are an author for Rampant TechPress (which you are, I am not implying otherwise).
In reviewing Tom's blog, we do find that Tom writes "I came up with examples and explanations that paragraph by paragraph showed this and sent it off to the website, asking them to please retract the article." This is not a demand, but a request. As I have no reason to doubt Tom, I accept that he did in fact send examples and explanations. This sounds perfectly reasonable to me. This is not a 'demand' nor is it a violation of the First Amendment to the Constitution of the United States.
Tom did not say "You must email the editor and demand that the article be removed." He simply noted that a feedback link existed by 'hinting' about it. I cannot and will not speculate as to his intention for writing what he did. As he did not say what feedback we are to submit, each person reading that entry is free to submit (or not submit) feedback that could be in support of the original article.
For an excellent example of how to handle public (or even private) criticism, I refer you to Jonathan Lewis's article on 'Reputations'. </code>
http://www.jlcomp.demon.co.uk/reputation.html Or you can read how Tom handles it as well (as he mentions an interaction between himself and Jonathan in the section headed "In closing...").
http://asktom.oracle.com/pls/ask/download_file?p_file=3067171813508366601
I have had the pleasure of interacting with Tom on several occasions over the past 2 years. He has always been a gentleman, very respectful of others and I have not heard him insult or demean another person. He can be blunt at times (he discloses this on AskTom) and he may use strong terms when disagreeing on a technical or professional topic. I have never seen him engage in an ad hominem attack when discussing a technical or professional issue.
http://www.nizkor.org/features/fallacies/ad-hominem.html <code>
In the past year, I had reason to question one of his entries on AskTom. His response to me was very professional, appropriate and to the point. The essence of his response was that the situation I was addressing had different 'parameters' and, as such, his advice did not really apply. I have no doubt that if I were to find what I think was a mistake in something he said or wrote, I could provide examples, logic and other evidence to him with my disagreement and we would have an appropriate exchange (regardless of the final outcome).
Just a BTW Harry
Bill S., September 13, 2005 - 11:46 am UTC
Harry, if you REALLY want to know why Gabriel removed his comment, why not e-mail him and ASK? If you click on his name in that removed comment, you get his blogger profile, where you can ALSO click on his e-mail address to send him a note. Just thought you might want to stop worrying about why he removed his comment.
For Harry
A reader, September 13, 2005 - 11:55 am UTC
Don and co should be very careful when they publish articles. It may not be your intention to mislead people (Nobody would do that).
But mistakes should be accepted and credit should be given to the person who pointed them out.
Again if you feel there are wolves trying to prey on you, why give them half a chance to do that?
Alex, September 13, 2005 - 12:02 pm UTC
I agree with Bill S 100%. Apparently all the folks at BC go through the same training course on how not to support your case with evidence before coming a member. They all say the same nonsense like a broken record.
All it takes is a cut and paste from sqlplus and show what your are trying to say. If they did that, I'm certain Tom would happily respond in a respectful manner they could have an intelligent, polite discussion instead of bickering.
Bob B
Bob B, September 13, 2005 - 12:06 pm UTC
Harry,
Show some proof that the tip works. Either an "Aritificial Test Case" or a "Real World System". If you choose the latter, I would expect to be able to put the hint into place and see the changes described and would ALSO expect that undoing the changes would undo the changes described. The only proof we've seen so far is a (mis)quote of Johnathan Lewis that said the hint was "substantially correct" ... in the following ways:
1) There are undocumented parameters.
2) Changing undocumented parameters can affect workarea sizes.
Careful Bob!
Bill S., September 13, 2005 - 12:14 pm UTC
That should read the "original" tip - exactly as it was when the very first iteration of that article appeared. Because THAT is what this whole thing has been about.
David Aldridge http://oraclesponge.blogspot.com, September 13, 2005 - 12:34 pm UTC
"... in the place I work we need answers now, not when someone has the time to answer. We also need answers that don't involve thousands of dollars of consulting or many days work
rewriting code."
So, what you're looking for is advice that:
* is of high quality
* is fast
* is cheap
* requires almost no effort to implement on your part
Not too much to ask then ...
Harry Conway - wake up!!!
Gabriel Paulovic, September 13, 2005 - 1:20 pm UTC
WOW!!! I'm famous!!! I'm a subject of an asktom discussion!! :-) ... ok, not that famous as someone (Harry Conway)has trouble understanding that when I add a comment I can also remove it. You don't like it? Sue me! ;-)
For Tom and others: I have removed the comment because after posting it I had noticed that the same point was already made (bug in metalink documentation) and that Tom already submitted tar/whatever to get it corrected. That's all. I just didn't want to duplicate the same info.
A reader, September 13, 2005 - 1:31 pm UTC
I read somewhere that you can get any 2 of the following for a piece of software ... it seems to apply to advice as well:
1) fast
2) cheap
3) quality
Pick any 2.
... and ...
Gabriel Paulovic, September 13, 2005 - 1:36 pm UTC
Harry, if you though my comment was in support of Don, I hope this will make it clear:
"It's not just Don who is WRONG ..." implying that also the metalink is wrong.
And I'm happy to report that metalink 223299.1 has been fixed. I don't think we can claim the same for Don ....
final words
G, September 13, 2005 - 1:47 pm UTC
Please Close Thread
Sathish, September 13, 2005 - 1:52 pm UTC
Tom,
Please close this thread. As someone already mentioned, its becoming like "predictive reorg".
Your valuable services are needed at other places..
Threads Don't Get Closed Here
Warren, September 13, 2005 - 2:03 pm UTC
I can see your point about the amount of input into this topic, but one of the great things on Tom's site is threads don't get closed.
At other sites, where the author(s) may not like what is being said, can't back up their comments, or are concerned about exposure threads get closed or even deleted.
Here, threads stay open and can be commented on by anyone who may be interested. You choose the model you like.....
Publish and be damned.
Jonathan Lewis, September 13, 2005 - 2:13 pm UTC
Harry Conway.
One of the problems of proposing a hypothesis that can easily be tested is that someone will test it.
When Tom asked you to supply any references to books which had been given one-star references you declined to offer any evidence, saying:
"There are plenty to choose from and they are easily spotted because the book review dates correspond to your condemnations."
The fact that you decline to support your claim despite the fact that it would be very easy to do so suggests that your claim cannot be supported. Moreover, you are encouraging people to dismiss your claims as 'typical Burleson bluster' - WHETHER OR NOT you are telling the truth.
Well, since you won't offer evidence, I did a quick check on Amazon.com - it's very easy to query all books from a publisher. Of the 62 titles under Rampamt Tech there are two which have received one one-star review each since Jeremy Smith suggested to Tom that Tom should give all the books one-star reviews. (You may note that Tom has not followed the advice - and has stated a couple of times in the blog that he reviews content, not publishers).
Possibly Jeremy Smith's suggestion was prompted by some form of deep disappointement with the book he had purchased, possibly the other one-star review was prompted by Jeremy Smith's suggestion, on the other hand it said:
"If you enjoy spending an open ended amount of time trying out as many tricks as possible and seeing what sticks without any understanding of what you are doing or how long the effect will last then this book may be for you"
so it seems possible that the reviewer had read at least part of the book.
Assuming a maximum of TWO reviews - that's not 'plenty'. And given that one of the reviews was from Jeremy Smith himself, I don't think you can even claim the luxury of two reviews.
In fact, your hypothesis is about as accurate as claiming you've implemented a '5x large increase' when you've provably done no such thing.
In passing - would you care to list a couple of the things in the original article which you thought were correct.
Can it be anymore funny than this....
A reader, September 13, 2005 - 2:13 pm UTC
I just looked up Mr Harry's resume on the Burleson Consulting site.
</code>
http://www.dba-oracle.com/resume_harry_c.htm <code>
It says:
"As a trained killer, Harry possesses those special skills required by BC to deal with obstinate consultants and end-users."
Don't know what it means! I never knew that such special skills are needed to deal with obstinate consultants and end users!! Tom, better you watchout...
Sigh
Bob B, September 13, 2005 - 3:31 pm UTC
One thread on Ask Tom has been closed to future comments(search for "Predictive Reorganization"). It wasn't closed because of a storm of technical arguments. It was closed because the commentary was getting out of hand with personal attacks (as is this one)
This thread is for discussing: Undocumented parameters, super sizing your pga, etc. On that note...
1) What are some ways to super size the PGA?
2) What are the pros and cons of super sizing the PGA in that way(s)?
3) When would it be worth considering super-sizing the PGA?
September 13, 2005 - 4:30 pm UTC
In a serial session - I would definitely prefer using manual memory management and setting sort area size/hash area size to what you want. If you want to override, then override and use the documented supported way to do it.
In a batch job
alter session set workarea_size_policy = manual;
alter session set sort_area_size = .......;
alter session set hash_area_size = .........;
For the parallel stuff, to make it larger, we'd have to take a look at it on a case by case basis. First, making sure it is even necessary - and second making sure the global settings you are making are reasonable "24x7" - for everyone.
A reader, September 13, 2005 - 3:45 pm UTC
Hear Hear! I agree Bob....
Bill S., September 13, 2005 - 4:26 pm UTC
[quote]Bob B Said:
One thread on Ask Tom has been closed to future comments(search for "Predictive
Reorganization"). It wasn't closed because of a storm of technical arguments.
It was closed because the commentary was getting out of hand with personal
attacks (as is this one)[/quote]
Folks, let's lay off the personal remarks please. I for one would LOVE to see Harry reply to Jonathan's post, as well as answer Tom's questions. You can certainly reply to a post, but let's leave the name-calling out of it if you don't mind.
And thanks!
This is the ONLY time when itÂ’s right to ask for removal
Harry Conway, September 13, 2005 - 5:08 pm UTC
I have to take exception to where asktom published:
"The Utilities book has NOTHING in it that is not in the Oracle Utilities guide."
I donÂ’t believe that an ethical organization like Oracle allows people to publish damaging statements of fact without verification on an Oracle-owned publication. I know that you moderate asktom, and that your remove what you believe to be inappropriate contributions to your publication.
- Tom, do you respect the rights of other people not to be unfairly smeared?
- What is OracleÂ’s policy on removing contributions when they can be clearly proven to be wrong?
- Is asktom a free-for-all for libel, or do you impose rules of fair play?
Let me anticipate your challenge to PROVE IT. Dave goes WAY beyond the Oracle Utilities Guide, which you should know if you had even read the title of the book “Oracle Utilities: Using Hidden Programs, Import/Export, SQL Loader, oradebug, Dbverify, Tkprof and More”: </code>
http://www.rampant-books.com/book_2003_1_utils.htm
Tom, you FREQUENTLY challenge your readers to PROVE IT, and I’m asking why you leave this clearly demonstrable falsehood on asktom. This sure looks like selective myopia to me. By the way, Tom, is this “Ryan” the same fellow who lives in your hometown of Leesburg, by chance? Does the IP address resolve to Northern Va?
If you bother to look, Mark Rittman likes the Oracle Utilities book and he notes Moore’s treatment of hidden Oracle utilities: ‘I also found Dave Moore's "Oracle Utilities", a look at the lesser-known and hidden Oracle utilities such as oradebug, dbverify, tkprof and Trace Analyzer an interesting read’
http://www.rittman.net/archives/2004_11.html
Here is more proof, an extract from the book where Moore shows how to analyze a tkprof report. ThatÂ’s NOT in the Oracle Utilities Guide:
http://www.praetoriate.com/teas_prae_util5.htm <code>
Well, there, I PROVED IT. Why didnÂ’t you, Tom? Was this just sloppy moderation, or do anonymous contributors have the right to unfairly defame people on asktom? Did you notify Dave Moore at www.oracleutilities.com ?
LetÂ’s see if you do the right thing. I think you owe Rampant TechPress and Dave Moore a removal, retraction and a sincere apology. If you donÂ’t, well I guess thatÂ’s more evidence that Janet Burleson was right.
September 13, 2005 - 5:29 pm UTC
If you feel the opinion of a previous poster is inaccurate, you are invited, encourage, allowed to post your rebut as you have.
I haven't read that book, I do know David Moore. But I have not looked at the book. If one person that read it got that impression, I believe they are allowed to state that opinion.
Thanks for the links to back up your claim, I think you've accomplished your goal. Now people have both sides of the coin and make their own decisions. The reviews on amazon are mostly favorable, David Moore is a bright guy - one person was apparently not impressed.
That is actually how the free press works.
(actually, how to read and analyze a tkprof is something documented
</code>
http://docs.oracle.com/docs/cd/B10501_01/server.920/a96533/sqltrace.htm#1018 <code>
and the link you provided does not really show us anything)
I do not actively moderate this forum. People are free to express opinions, provide reviews.
Confusion here Harry -- how could doing what Janet says is horrible to do (remove stuff cause someone said so) here be doing what Janet says to do?
A bit of tail chasing and circular logic going on there.
We all get bad reviews from time to time - not everyone likes everything I write. Not everyone will like a particular book.
And why do you get to decide when the ONLY time something is right is?
How about some documentation on why the seetings are so good?
CancunKid, September 13, 2005 - 7:42 pm UTC
Mr. Kyte,
You seem to be arguing that the undocumented parameters are set so wonderfully that they should never have to be touched. Where is all the documentation to show the research Oracle did to determine that every one will only need a maximum of 200 megabytes for hash and 100 megabytes for sorts? This seems to be rooted in the mistaken belief that we will always have to be stuck on 32 bits (why else have 4g the upper limit?) Why 5%? Why 30%? Why have the undocumented settings at all if they are not there to adjust the way the sorts and hashes are done? Or was the research a bunch of developers and a dartboard?
I have read Mr. Lewis and Mr. Rogers as well as everything you have said. I do not see anything proving the "dangers" that you so ardently claim exist of setting these parameters. Instead I see proof they do as claimed, allow sensible setting of parameters Oracle seems more to have guessed at.
September 13, 2005 - 8:14 pm UTC
where do you see that? I'm stating that the article was totally wrong and is now only partitially wrong after incorporating many of the comments made about it.
The issue we have is the article was entirely inaccurate in describing these parameters, how they work, how PGA memory management works.
You are trying to 'move the proverbial cheese' CancunKid, this was about *a bad article*, not about the merits of automatic pga memory management (you see, that article wasn't about pga memory management - at least not Oracle's implementation that is - see Jonathans first article about it, the title sort of sums it all up).
It was a bad article about *undocumented* (look up that word, it means - well, undocumented, asking for the documentation for that which is undocumented is sort of, well - you fill in the blank)
Do you realize the article *DOWNSIZED* the pga for parallel sorts and *DID NOT AFFECT* the size of a serial sort at all when it was written?
And this a paper entitled "super sizing", do you think the author that wrote:
<quote>
For example, the following set of parameters indicates that the Oracle DBA does not understand PGA management.
* sort_area_size=1048576 <-- sort_area_size is ignored when pga_aggregate_target is set and when workarea_size_policy =auto.
* pga_aggregate_target = 500m <-- The maximum default allowed value is 200 megabytes
</quote>
might have been the one with a lack of understanding?
How could an article that says "set these parameters this way and you'll see a 5x increase in memory usable by the system!!"
But actually shows settings that don't affect memory at all - or restrict the amount of memory you can use?
<quote>
the example in Figure 1, the following settings would increase the default sizes for large sorts and hash joins.
* pga_aggregate_target = 1000m
* _pga_max_size = 1000m
* _smm_px_max_size = 333m
With these hidden parameters set we see a 5x large size increase for parallel queries and sorts:
* A RAM sort or hash join may now have up to 50 megabytes (5% of pga_aggegate_target) a 5x increase.
* Parallel queries may now have up to 330 megabytes of RAM (30% of pga_aggegate_target), such that a DEGREE=4 parallel query would have 83 megabytes (333 meg/4).
</quote>
the problem is, with or without the magical _ parameters -- having the pga_aggregate_target at 1000m was sufficient to permit workareas of up to 50m ALREADY. The setting of the parameter -- did nothing there.
And as Jonathan demonstrated -- setting the _smm_px_max_size down sized - yes, decreased, reduced, limited the amount of ram.
Interesting?
So, how does that "allow sensible setting of parameters Oracle seems more to have guessed at"
Who pray tell might have been doing the guessing here? I wonder.
That last part is really funny, our developers just throw darts at boards - yup, you got it. I'll bet they even close their eyes when they do it!
Someone must sometime explain the 5x increase with settings that *decrease* the memory -- do that for me, would you??
Had the article
a) correctly and explicitly defined the undocumented parameters
b) demonstrated a clear understanding of how they work
c) demonstrated a clear understanding of how automatic PGA memory management
works
d) actually gave a tip (that is the kicker, isn't it? The "tip" in a paper to
super size, down sizes?)
I would not have had any problems with it whatsoever.
Now, if you would like to dig up the article and actually discuss its accuracy, oh wait - Jonathan did that already.
So, Cancunkid, what is your point? Is incorrect, inaccurate, misleading information something that should be just left out there? Lets say you were the DBA with sort_area_size set and pga_aggregate_target at 500m and someone said to you "you daft sorry excuse for a DBA, this paper by a well known name says you dno't understand how to do your job".
Is that
a) a good thing
b) a bad thing?
CancunKid
Gabriel Paulovic, September 13, 2005 - 8:08 pm UTC
CancunKid, if running an unsupported version of Oracle (unless you were instructed by Oracle support to set the udocumented parameter you are running an unsupported version of Oracle) does not constitute a danger for you then, well ... you just don't need the data, ... and databases are, above all, about the data, not about a cool ways to configure database parameters.
Furthermore, I do not see Tom suggesting that you might not need larger sort area size. But if you need it there is a way to do that WITHOUT resorting to undocumented parameters. I don't get it why you think you HAVE TO use pga_aggregate_target/ workarea_size_policy = auto and fiddle with undocumented parameters when you can just set workarea_size_policy = manual.
And if you REALLY THINK you need that undocumented parameter set ... well ... see Oracle support.
September 13, 2005 - 8:21 pm UTC
(and you'd really want to read an article that actually showed how to INCREASE them, not DECREASE them wouldn't you??)
Yes, I think that is dangerous "lets do this, it is so cool, and he says it will increase the memory we use".
Little knowledge (of how things actually work) is in fact a dangerous thing
A reader, September 13, 2005 - 8:16 pm UTC
"CancunKid from NJ" atleast you learned one good thing from this site is asking for proof.
CancunKid
A reader, September 13, 2005 - 9:05 pm UTC
What kind of test have you done to find out the "maximum of 200 megabytes for hash" ?
Also as Gabriel Paulovic said, if you really need this parameter just check with support.
David Aldridge http://oraclesponge.blogspot.com, September 13, 2005 - 10:32 pm UTC
With regard to setting these in a batch job:
alter session set workarea_size_policy = manual;
alter session set sort_area_size = .......;
alter session set hash_area_size = .........;
I think that I'd be tempted to have a table of values for these parameters to be read and used by particular batch jobs -- so if you want to tweak 'em you don't have to hack the code. Just a maintainability thought.
September 14, 2005 - 7:55 am UTC
absolutely. If desired, they do not have to be hard coded.
Why min(5%,100MB)
VA, September 13, 2005 - 11:45 pm UTC
See Metalink Note 147806.1
<quote>
PGA_AGGREGATE_TARGET limits both the global PGA consumption and the size of a workarea i.e. the memory allocated to a single SQL operator is also limited to min(5% PGA_AGGREGATE_TARGET, 100MB) for serial operations, and to
30% PGA_AGGREGATE_TARGET/DOP for parallel operations DOP=Degree of Parallelism)
</quote>
Why min(5%,100MB)? Heck, why 5%, why 100MB? These days, servers with 64GB of RAM are fairly common. In such servers, it is not inconceivable to set pga_aggregate_target to, say, 4GB or more. Due to the above "design", "memory allocated to a single SQL operator" would be limited to 100MB.
What is the rationale behind this design decision?
Thanks
September 14, 2005 - 8:10 am UTC
the pga_aggregate_target is a system wide target.
If you were able to take 100% of it, think about what the consequences would be.
You - your single session could consume all of the memory for your first sort, leaving nothing for anything else.
Since you can and will have more than one workarea (think about two queries, both did sorts and both are opened, you have two work areas *at least*, if not more).
the pga automatic stuff is to enable the sharing of a resource, therefore, you do not give 100% of the resource to the first person that asks for it.
And it is not the min(5%,100mb) - various operations will take a percentage of 5% and 200mb in a serial operation.
You can read more about it here:
</code>
http://oracle.apress.com//betabooks/betabook.html?betabook_id=30&APRESSESSID=c5e74fa001f64357db462f5e677af1b1 <code>
(if in the future that link is dead, it is because the book was published and they took down the beta chapter)
But, if you don't like the auto stuff you are free to turn it off and go manual. Just like with automatic undo management, automatic sga management, automatic <anything>.
Harry - why not answer the questions?
Bill S., September 14, 2005 - 12:00 am UTC
Harry, you're just trying to move the cheese (or hide it behind a smoke screen). You STILL haven't answered Jonathan's questions, or Tom's either. Instead, you continue to post baseless accusations in an attempt to deflect everyone's attention from the real issue. Your frivolous statements in your first post still haven't got one shred of supporting evidence anywhere.
So Harry, how many times are you going to come back and post yet another bunch of ridiculous rumours, and avoid answering the questions? If that's all you're going to do, you may as well quit now - you're down by THREE now, and I don't think there is any way you can come up with a winner.
You claim the original article was accurate as published.
You claim that Tom FALSELY stated that it was "just wrong".
You claim that Rampant books are being given undeserved one-star ratings in DROVES.
You claim a lot of things. So when are you going to provide anything of substance to support these claims? Or are you just going to continue wasting bandwidth by posting ridiculous charges?
Is it true?
A reader, September 14, 2005 - 1:37 am UTC
Ryan wrote this fact: "The Utilities book has NOTHING in it that is not in the Oracle Utilities guide."
Only a moron would suggest that his was an opinion. It is clear, unambiguous and as you say Tom PROVABLE.
So, is it true or not?
I saw your blog and it was a public beating, without any justification whatsoever.
I never had much respect for your books anyway, but I'm not going to sit back and watch this load of poo.
You remind me of the schoolyard bully . . . You are a disgrace to Oracle, Mr. Kyte.
September 14, 2005 - 8:13 am UTC
Have a nice day.
Have you posted enough drivel?
A reader, September 14, 2005 - 1:53 am UTC
Tom, you posted this drivel in two separate threads and now here, on your employers own forum.
Same crap, same demands to slam Don. Why three separate threads, Tom?
Maybe because the first two did not cause enough harassment to have the tip removed?
It has about a 4 out of 5 before you started this, and you only managed to slam it down to 2.5.
Clearly not everyone disliked this tip, and you are beyond offensive to suggest that it be removed.
Did you read this?
</code>
http://dba-oracle.com/oracle_news/2005_9_8_blogging_job_fired.htm <code>
I'm a stockholder, and I have something to say about this, but I can't put it here.
What is the Oracle blogger policy Tom?
September 14, 2005 - 8:23 am UTC
Maybe because someone came here and asked a simple question (did you read the question? it is at the top of the page).
If you care to point out anything specific you feel was baseless (as in "not truthful", "not grounded in facts"), I'd be glad to look at it.
Otherwise you are just spouting off.
No demands to slam anyone. Just voicing my opinion on the damage factually inaccurate material causes.
Whether people liked or disliked the tip isn't even relevant. Think about it - the tip did not work, it claimed to SUPER SIZE (after slamming DBA's for not understanding pga memory management) when it really DOWN SIZED.
Now, if you "liked" that tip - perhaps it could be because you trusted the author, but the author let you down (and you had not yet realized it).
So, you might have "liked it", but if you use it, you will let down - as it did not do what it proposed to do.
and then you look "not so smart".
Maybe they would like it less had they had the facts?
To the reader: Come out will you!
Saikat Chakraborty, September 14, 2005 - 3:35 am UTC
<quote>
I never had much respect for your books anyway, but I'm not going to sit back
and watch this load of poo.
</quote>
Well, nobody is perfect and how can I blame you for being mentally incapable to have a respect for books that prove what it teachs?
But at least come out of the assumed names while attacking a person who writes books and helps much more persons than you can hiding behind your assumed identity. By the way are you one of "the fastest-growing Oracle book publisher in the world"?
Hear Hear
Jim, September 14, 2005 - 5:09 am UTC
<quote>By the way are you one of "the fastest-growing Oracle book publisher
in the world"? <Quote> Oh, I think so ;) either that or a relation/buddy
Come on Guys? Be professional?
Tony, September 14, 2005 - 6:00 am UTC
Lovers of Don?
Please don't spoil this thread. If you love Don, consult him, employee him or whatever... My experience with Don's book is worst, so I stopped not even buying his books but even hesitate to recommend to anyone. I like Tom's books, and I wait for them. Please go to Don's site, and appreciate there, he needs lots of people there. I hope you will stop childish behaviour here. If you continue to do this, time is not far when some body will sue Don for spreading wrong knowledge, and ask for compensation...
Thanks
A couple of comments
Niall Litchfield, September 14, 2005 - 8:21 am UTC
A Reader wrote "Ryan wrote this fact: "The Utilities book has NOTHING in it that is not in the
Oracle Utilities guide."
Only a moron would suggest that his was an opinion. It is clear, unambiguous
and as you say Tom PROVABLE.
So, is it true or not?"
Did you read the exchange with Harry Conway in this thread? You know the one where Tom welcomed fact based rebuttal of Ryan's assertions. I think this one has already been answered.
A.N.Other Reader wrote
"Tom, you posted this drivel in two separate threads and now here, on your
employers own forum.
Same crap, same demands to slam Don. Why three separate threads, Tom?
Maybe because the first two did not cause enough harassment to have the tip
removed?"
Maybe because he was asked a question about it?
"It has about a 4 out of 5 before you started this, and you only managed to slam
it down to 2.5.
Clearly not everyone disliked this tip, and you are beyond offensive to suggest
that it be removed."
Now that is an opinion I can't agree with. The article was wrong. It is now less wrong, as a direct result of the discussions on Tom's blog and Jonathan's articles. That is, the suggestion that it was just wrong and the demonstration that it was just wrong have improved the article, somewhat at least. Would it actually have been better to leave the original article there?
"Did you read this?
</code>
http://dba-oracle.com/oracle_news/2005_9_8_blogging_job_fired.htm" <code>
I have, its all a bit speculative and woolly isn't it? Tom's blog doesn't seem to break any of the guidelines listed in the article either, whereever they come from. I'm not sure about the North Carolina link (its broken) - certainly if publishing false information is an offense in North Carolina, I'd be worried a bit if I was based there and information that I'd published were shown to be false - I suspect it has to be rather more than just wrong though, the newspapers would be empty otherwise. The link that purports to show what English law is (not British by the way, British law doesn't really exist) actually shows someones interpretation of what the courts might hypothetically decide here. I'm told that the Australian law is not as represented there either.
Harry "Con" way
Mark Evans, September 14, 2005 - 8:34 am UTC
Harry, do you have any idea just how sad and pathetic you’re sounding? And the rest of the BC clan? What’s so wrong with Tom and Jonathan and others correcting an article by your boss Burleson that’s so full of obvious errors? Did you really believe the ludicrous notion that the pga_aggregate_target is limited to 200M? Did you really believe the ludicrous notion you need undocumented parameters to set an effective pga_aggregate_target over 200M? Did you really believe the ludicrous “real life” scenario by Ault, now sadly gone, you need to set undocumented parameters to avoid 64M one-pass operations? These are just a tiny, weeny sample of the glaring, dangerous errors in the articles that suggest Burleson has little idea on what he was writing about and yet you claim “the original article looked fine to me”! Why then all the subsequent re-writes? Harry, if you still claim the original article looks fine, it’s not really a great reflection on your technical expertise is it, not the best way to promote yourself as an authoritive author? So Harry, why is Tom so wrong with his assessment of your boss’s article? Where are Jonathan’s technical errors with his comprehensive “demolition” of Burleson’s article, which incidentally is not quite so complimentary as you suggest?
I donÂ’t expect any answers because you seem incapable of providing anything other than wild accusations and innuendo. Where are all these 1 star reviews Harry? There are lots and lots and lots of 5 star reviews by fellow Burleson Consulting employees and Rampant authors, but not too many 1 star reviews since all this began? YouÂ’re right, there appears to be some funny busy going on but not how you suggest.
Listen Harry, do you hear that buzzing noise whizzing past your shoulder? ThatÂ’s your credibility flying out the window.
Mark
Harry's response is fair
Ryan, September 14, 2005 - 9:12 am UTC
I mispoke. I should have said it didn't have anything in it that you can't get out of the documentation in general. I did not read the book word for word. The parts that I read were pretty much right out of the utilities guide. If I see that why would I read the rest? I am sure most people do this. You go to the bookstore, you flip some books and see what is in them. If I see material that I can get for free, I see no reason to purchase the book.
The part about demanding a review removed is silly though. The rampant press crew routinely review each others books and say they are terrific. A negative review is not the end of the world. The sun will come up the next day.
September 14, 2005 - 9:34 am UTC
I agree, Harry posting of his point of view is totally fair.
Demanding that an opposing point of view be removed though?
But frankly, I would say you cannot judge a book by glancing at it like that. If you look at my book - the outline looks a lot like the concepts guide outline.
However, I believe you would find upon reading it to be very different in the way it relates the information.
And sometimes, that is what it is all about - how the information is related.
And I thought you guys could do better than that.....
Bill S., September 14, 2005 - 9:28 am UTC
Funny, the WHOLE point of this thread (see the beginning) and Tom's blog entries AND Jonathan's articles was......
that the ORIGINAL ARTICLE WAS DEMONSTRABLY WRONG.
Say it three times, if it helps you. So some folks that would like to support Don have come here and begun posting. They post that Tom is unfairly harassing Don; that the ORIGINAL ARTICLE was fine; that comments by people posting to Tom's blog (and here) are Tom's responsibility.
Nice attempts to deflect, BUT WE HAVE YET TO SEE ANY EVIDENCE THAT THE ORIGINAL ARTICLE WAS CORRECT. I've seen tons of evidence showing how wrong it was. Not one spec of where it is completely right. Not one.
Tom has been trying to get that article corrected since day one. The author apparently didn't want to correct it, the editor who published it didn't want to retract it. So Tom wrote a blog or two about it. Jonathan Lewis wrote several articles refuting it. And it magically gets "enhanced" not once but twice;but still folks say it was fine the way it was. Everyone has opinions, and they are entitled to them. If something someone posted, either here or as a comment on one of Tom's blogs, is DEMONSTRABLY WRONG then feel free to post the refuting evidence. No-one will mind. We'd actually all WELCOME IT. If you can't provide the evidence to back up your statements, maybe (perhaps, just maybe) you shouldn't be making them?
"A Reader" (just a few posts ago said [quote]Ryan wrote this fact: "The Utilities book has NOTHING in it that is not in the Oracle Utilities guide."
Only a moron would suggest that his was an opinion. It is clear, unambiguous and as you say Tom PROVABLE.
So, is it true or not?[/quote]
Well, I haven't read that book - have YOU? Is there ANYTHING in it that isn't in the utilities guide? If Ryan feels that way, fine. IT IS HIS OPINION, and HE IS ENTITLED TO IT. It seems to me only a moron would suggest that it WASN'T Ryan's opinion since he so obviously posted it. Book reviews are most certainly OPINIONS, shall we drag out a dictionary for you "READER"?
Another "READER" (probably the same one, I think) said
[quote]
Clearly not everyone disliked this tip, and you are beyond offensive to suggest that it be removed.[/quote]
Gee, it was never a question of whether or not someone LIKED the tip, it was about the tip being DEMONSTRABLY WRONG. If you think it offensive to request the removal of provably incorrect material, well...I just don't know what to say to that. Seems to me you "READER"s are up on your toes screaming to have FACTUALLY CORRECT material removed - weird world, ain't it? And regardless of what else has been said, you can feel free to post anonymously - we don't care about that. Just try to post something that at least RESEMBLES facts. With supporting evidence. That's what we expect to see from professionals.
Alex, September 14, 2005 - 10:28 am UTC
I think the folks at BC need to ask themselves a few questions:
Why would Tom and John go through the trouble of writing tests and articles correcting Don's work? Is it because they are concerned people might use the false information and cause themselves headaches?
Or is because they're jerks and they're trying to steal your business? I have news for you guys, Tom's book is going to fly off the shelves whether he critiques you or not. This is because he is outstanding, and not for any other reason. Time to swallow your pride and just say, "thanks for pointing that out, well done." I've done it about a 1000 times for each of the times Tom's corrected me.
Don's crew is digging their hole deeper
jim, September 14, 2005 - 10:44 am UTC
I am amazed that Don's crew would post such bluster here. But then Janet did so his minions are just following his example. Sigh, they would have been better saying nothing or just correcting the information and moving on.
Their lack of professionalism is surprising. You can be sure I won't be hiring them, recommending them for a consulting engangement, or buying their books. Is that unfair? Perhaps, some of the books might be good, but as Tom had on a seperate blog, it can take a lifetime to get a good reputation, but moments to destroy it. Don and crew's behavior have destroyed any good reputation they had.
September 14, 2005 - 11:34 am UTC
But don't disregard all of the authors, really - that would be unfair. Both to you as well as the author themselves.
I won't even think about writing a list, I'll just use one example, one that I've used before (there are many others).
</code>
http://www.oracle.com/technology/community/oracle_ace/ace4.html#nanda <code>
Arup Nanda
I've known him a long time, I know what he does for the Oracle Community. I know his work. In my opinion, if he wrote it - it is likely accurate and useful. He is careful in his work, he "does it" himself (he is an Oracle DBA, day to day).
I would hate to see everyone clumped together.
My take on this
A reader, September 14, 2005 - 12:03 pm UTC
Here is how I see this:
1 – Burleson published an option piece titled “Evaluating the CREDIBILITY of Oracle information on the web” that included a fair comment about Tom Kyte’s policy to “make fun of” his readers if they do not communicate properly with him.
2 – Tom waits for the next Burleson article on SearchOracle.
3 – Tom Kyte attacks it, using the titled “Getting CREDIBLE information”. See the pattern here? Tom know from previous experience (The infamous “predictive reorg” thread), that he will get loads of readers to do he dirty work.
4 – Tom Kyte starts yet another blog asking “why does this article still exist?”
5 - Tom is still not satisfied and starts this thread, on the exact same topic.
For the record Mr. Tom Kyte, Oracle employees DO NOT have the right to “make fun of” people on an Oracle forum, and I applaud Burleson for exposing your unprofessional and shameful policy.
Let’s call this what it is. It looks like it had nothing to do with the tip, it was just Tom Kyte getting even for being exposed. Also, your “I prove it wrong” nonsense applies to the Oracle university courses where they omit exceptions to make it easy to learn.
Why, Tom Kyte, didnÂ’t you ask Oracle University to remove their classes?
September 14, 2005 - 1:27 pm UTC
Well, thanks for not using instant messaging speak :) Appreciate that.
You do understand there was slightly more than one article between then and now right? I mean you are going back to March 2005 with the credible thing. Why would I lay in wait so long?
But here is how I view it, it doesn't take a google search, it doesn't take "not making fun of", getting credible information has to do with, well, getting credible information - credible as in:
credible Audio pronunciation of "credible" ( P ) Pronunciation Key (krd-bl)
adj.
1. Capable of being believed; plausible. See Synonyms at plausible.
2. Worthy of confidence; reliable.
and for that, I have one way to judge it - it isn't based on who wrote it, why they wrote it -- it is based purely on "it is correct"
And this article was so far from correct (you all keep glossing over that - this was not "unwarranted", this article was so wrong as written) that something had to be said.
Who says what Oracle employees have the right to do or not to do? It isn't like this site is hiding anything?
How do you know what the motivations are? You want to know the story behind the story here?
o Ok, I recieved an email - asking me about that web page (someone said "hey, is this true").
o I read the web page.
o It was pretty much "wrong"
o So, I wrote an email to search oracle with 8 points -- stopping there and suggested they take it down.
o In order to make it not just "be me saying so", I asked Jonathan to take a peek, he came to the same conclusion "wrong stuff there"
o He published his article, the same day I did mine.
Exposed? How have I been "exposed" -- you did sort of read the page you used to fill in this followup didn't you????
It states very clearly:
FAIR WARNING
if you use 'IM' speak (you know, u wrt like this w plz and thnks), I will make fun of you. I will point out that it is not very professional or readable.
FAIR WARNING
In bold, on the top.
Now, if you needed a slueth to help you find that out, well........ Ok.
When I find wrong and incorrect material, I get it fixed. I don't see everything so I cannot fix everything.
One good of all of this -- some metalink notes got fixed, that was awesome.
Agree with previous post
A Tom Fan, September 14, 2005 - 12:09 pm UTC
i tend to agree with previous post. also to note, have you ever visited metalink? It went from bad to worse and then from worse to nowhere. If you get help from the forums, then its from other users. the only help you get from the support specialist is posting "thanks for using oracle forums". since tom represents Oracle, I guess he should start cleaning his mess before pointing elsewhere
The problem with getting into a mud slinging match with an idiot
Mike Friedman, September 14, 2005 - 12:15 pm UTC
You still get covered with mud.
I think it is clear to everyone that the Burlesons are not quite right in the head.
Leave it at that.
The truth hurts
mdinh, September 14, 2005 - 12:25 pm UTC
And I will leave it at that.
On CyberLibel
A reader, September 14, 2005 - 12:26 pm UTC
</code>
http://www.lectlaw.com/def/l032.htm
“Criminal Law. A malicious defamation expressed either in printing or writing or by signs or pictures, tending to blacken the memory of one who is dead, with intent to provoke the living; or the reputation of one who is alive and to expose him to public hatred, contempt or ridicule.”
FROM THE CANADIAN BAR ASSOCIATION:
http://www.cyberlibel.com/libel.html
“Does the Plaintiff have to prove the alleged libelous statements are true? No. The law presumes in the plaintiff's favour that the statement is false, unless and until the defendant proves the contrary.
Owners and/or operators of a network of computers are liable for defamatory material which they write and publish on the network, or receive from third parties and cause to be published on the network. Consequently, a corporation could be vicariously liable for statements published by employees.”
http://www.cyberlibel.com/cbaopupdate.html <code>
- The statement has been made to a third party.
- The statement referred to the plaintiff. This does not mean that the statement has to refer expressly to the plaintiff. A statement can be actionable if it is reasonably capable of referring to the plaintiff.
- The statement must be defamatory, which means that it must be a false statement to the plaintiff's discredit.
The law presumes that a defamatory statement
a) is malicious
b) is false
c) caused damage”
“Your Corporation's Response to Rumour: . . write and demand a retraction and apology”
I wonder if Harry “Con” way, (you know, the person that ASKTOM has published is an idiot), the one whose credibility is flying out the window, knows about these laws?
Does Oracle Corporation respect the laws against defamation in Canada, Mr. Kyte?
David Aldridge http://oraclesponge.blogspot.com, September 14, 2005 - 12:27 pm UTC
I do not understand why it might be OK to criticise Oracle University, Metalink, Oracle Support, or Oracle-supplied documentation for any failings or technical inaccuracies on their part, yet it is not OK to criticise other information sources for technical inaccuracies.
This is what the issue is about -- technical accuracy or inaccuracy. Nothing to do with vendetta, personal feelings or loyalties, egos, or even professional reputation. Once the bones have been picked out of this sorry business it comes down to whether an article gave accurate information on how to modify Oracle's behaviour in adjusting various components of the PGA memory -- nothing more and nothing less. The facts are clearly demonstrable, and there's nothing more to be said.
Telepathy, or small world
Alberto Dell'Era, September 14, 2005 - 12:32 pm UTC
I was reading Arup Nanda's "Oracle Database 10g: Top 20 Features for DBAs" (and the addendum for 10gR2) on OTN and enjoing his style, conciseness and accurate examples, when I found the Tom's post about him above. I concurr with Tom, it would be a shame if someone didn't read his book(s) only because the publisher is Rampant. It may be my first Rampant book actually.
Getting Personal
Dr. Steven M. Cerutti, September 14, 2005 - 12:38 pm UTC
I am a Rampant Tech author. I am not an expert on Oracle --I don't even know what Oracle really is. I am a Professor of Classics with a Ph.D. from Duke University and I teach courses on Greek and Roman antiquity. I know the entire Burleson family, have had both of Don's children in my classes, and am proud to be the first author of Rampant Tech's new "Pedagogue Series" that is going to expand Don's press into the academic world. What does this say about Don? That he doesn't care about education? And what is education, if not about disseminating knowledge. I have read these reviews and find them to be for the most part personally motivated. I have not read the "article" in question. I probably wouldn't understand a word of it. But I have worked closely with both Janet and Don in bringing forth this new academic arm of his publishing business, and know the kind of integrity Don and Janet Burleson possess. On a professional note, this will be my fifth book. I have worked with other publishers, and none of them rises to the standards of Rampant Tech. I'm sure there will be errors in my book, as I find errors in other books. All this seems to be about is not an inaccurate article, but a personal feud that sounds more childish than anything I have encountered in academia, and that's saying something.
Getting Personal
Peter K, September 14, 2005 - 12:55 pm UTC
I think to be fair to Tom. He has never gotten personal but rather various other posters including Harry Conway, Janet B and various "Readers" had.
For example, Tom was accused of starting this thread when in fact, it was a question posted on AskTom to which he provided a reponse.
If Don has the professional integrity as Dr. Cerutti suggested, then corrections to the article and an acknowledgement to the folks who pointed out those errors and provided the information should be stated but instead the article was "enhanced".
Like I had indicated on other forums. The number of folks to avoid at BC has climbed to 5. Would I avoid Rampant Press books? Only certain ones by certain folks.
Time to locked this thread
Peter K, September 14, 2005 - 1:02 pm UTC
Given that the postings have gotten very personal with a lot of various folks letting fly accusations and what not, I think (MY OPINION ONLY) that this thread on AskTom should be closed/locked or whatever else you call it.
Each of us have our own opinions and let's agree to disagree.
The fact of the matter is that there are a lot of incorrect information available on the Net and it's in our interests to read, test and apply whatever is applicable to our unique situation.
Lock this thread?
Doug Burns, September 14, 2005 - 1:18 pm UTC
Peter K,
On the one hand I agree with you - this is just a waste of everyone's time because these people clearly can't see their problem. Their responses are just bizarre and I can't see it going anywhere.
On the other, I think it would be giving in a little. Maybe I'm being paranoid, but all these Rampant Press authors and Anonymous 'readers' springing up strikes me as something of a concerted attack. Much more concerted than anything these guys claim is coming from Jonathan and Tom. (A clue for you guys - neither Tom or Jonathan is my boss, so I owe them nothing.) So I'd feel like I was giving in to a school bully.
At the back of my mind, I know people have been threatened via email, rather than in public forums, and I don't like it.
I'll expect to receive my own email shortly ...
P.S. Quick question for the Rampant guys. I went looking for these terrible reviews you were talking about and all I could find were glowing reviews from your co-authors. Aren't you embarassed?
Response to "A reader" - On CyberLibel
Gabriel Paulovic, September 14, 2005 - 1:23 pm UTC
<quote>
I wonder if Harry Conway, (you know, the person that ASKTOM has published is an idiot), the one whose credibility is flying out the window, knows about these laws?
Does Oracle Corporation respect the laws against defamation in Canada, Mr. Kyte?
</quote>
Let's be clear. Harry Conway was the one making FALSE accusation here not Tom. Harry Conway accused Tom of removing or contacting me to remove my comment from his blog although it CLEARLY was not the case. I'm witness to that and will gladly attend any court proceeding testifying on behalf of Tom. Yes, you are right, Harry's credibility is flying out of window ... because of his false accusations and unprofessional conduct in this discussion. He might be able to sue himself I guess. ;-)
Furthermore, I'm not sure why Tom would be concerned with any Canadian law, for that matter, I'm not sure why Harry Conway would be concerned with any Canadian law. Please, do not pretend do know a squad about any law.
David Aldridge http://oraclesponge.blogspot.com, September 14, 2005 - 1:38 pm UTC
Dr Cerutti,
While your testimonials on the good character of Don and Janet are not unwelcome, they surely can have no bearing on the accuracy of technical articles. Even if the criticisms of the articles were motivated by malice or personal issues, and I personally do not believe that they were, then that also would have no relevance to the accuracy of either the articles or the criticisms.
The judgement of the accuracy lies solely on technical proofs, and the publication of such proofs is both their strength and their potential weakness -- anyone is free to challenge them or to provide a counter-proof.
On the subject of books, I recommend "Crimes Against Logic" by Jamie Whyte.
Alex, September 14, 2005 - 1:45 pm UTC
Tom I think you may have missed something:
<quote>
Also, your “I prove it wrong”
nonsense applies to the Oracle university courses where they omit exceptions to make it easy to learn.
</quote>
Well there you have it, check and mate. Can't argue with that....That big long proof at the top of the page is just "nonsense". Since you put it that way....
Laughing out loud.
A reader, September 14, 2005 - 2:33 pm UTC
Gabrial lied: “Harry Conway was the one making FALSE accusation here not Tom. Harry Conway accused Tom of removing or contacting me to remove my comment from his blog although it CLEARLY was not the case. I'm witness to that and will gladly attend any court proceeding testifying on behalf of Tom.”
Good thing that you will testify, because you have just libeled the idiot, Mr. “Con” way. He did not accuse Tom of removing your post, and if you knew how to read you would know what he said:
“You also allowed one comment to be removed by Gabriel Paulovic, without any explanation. Tom, did you contact Mr. Paulovic and ask him to remove his comment?”
Gabriel, calling you stupid would be an insult to Village Idiots everywhere. Please apologize and crawl back into your hole.
Thanks David!
Bill S., September 14, 2005 - 2:36 pm UTC
David Aldridge said:
[quote]
This is what the issue is about -- technical accuracy or inaccuracy. Nothing to do with vendetta, personal feelings or loyalties, egos, or even professional reputation. Once the bones have been picked out of this sorry business it comes down to whether an article gave accurate information on how to modify Oracle's behaviour in adjusting various components of the PGA memory -- nothing more and nothing less. The facts are clearly demonstrable, and there's nothing more to be said.[/quote]
Yup. To use someone else's favorite phrase, it's self-evident and almost a tautology.
Still lots of words but no proof
CancunKid, September 14, 2005 - 2:53 pm UTC
So once again I ask...were did the mystical 5% come from and the setting of 200 megabytes? It seems to harbor back to the 4 gig limit in 32 bit Oracle since 5% of 4 gig would result in 200 megabytes. But what if I have the memory and all my users need more than that and I want automatic management? I am out of luck? Seems a bit cavalier If I want to also prevent the parallel queries from getting out of hand, what then?
Thanks
September 14, 2005 - 3:34 pm UTC
Why, we are not talking about the merits of the algorithms (and the 5% is for "sharing", it is this thing called "sharing" see - the auto stuff is trying to make the amount of ram available vary with different loads, different numbers of users -- if they didn't use 5% - the first user could take *it all*. What then - have you thought it out here)
But again, we were not discussing the merits of the existing, undocumented, automatic algorithms.
Rather we were talking about the factually inaccurate portrayal of how the software actually works.
You want to talk about undocumented cool stuff, at least understand it first.
Yes, I guess you'd be out of luck, especiall if you read that article
(remember Cancunkid - the premise of the article was -- this is a gas -- "super size", but it proceeded to show how to downsize -- keep that in mind)
A paper that showed how they worked
Gave use cases for them
Explained it as reality has it, not as made up
And did what it stated it did (super, not down, size)
I'd have no problem with that.
A paper that
Showed the author didn't understand how they worked
Didn't really give any use cases
Explained it in a manner not in keeping with reality, how it really workd
And did the opposite of what it stated it did
I have quite a few issues with that.
dear "A reader"
Gabriel Paulovic, September 14, 2005 - 3:04 pm UTC
<quote>
Gabrial lied: “Harry Conway was the one making FALSE accusation here not Tom. Harry Conway accused Tom of removing or contacting me to remove my comment from
his blog although it CLEARLY was not the case. I'm witness to that and will gladly attend any court proceeding testifying on behalf of Tom.”
</quote>
Harry Conway accused Tom of removing OR contacting me to remove ... can you see the "OR" part? ... the true is I have said before that Harry accused Tom removing my comment but I realized that MIGHT not have been the case. That is what happens when English is your second language.
So let's clarify: Harry Conway FALSLY accused Tom of asking me to remove my comment. Happy now?
You might at least have a courtesy of spelling my name correctly when you are trying to insult me. It is GABRIEL.
Harry, Don or other BC employee or associate, did you change also you legal name to "A reader"? :-)
How do you write it
A reader, September 14, 2005 - 3:08 pm UTC
I have a highest regard for your abilities and I hoping that you might show how you might have written this tip yourself. Like Don tryed, talk about your own experience using hidden PGA with customers and share storries of what happened.
I donÂ’t want scripts, I want to hear your experiences please.
September 14, 2005 - 3:50 pm UTC
I did, it is a chapter in my book, I've put quite a few pointers to the chapter on how memory is managed by Oracle.
I don't use hidden parameters - virtually every time, they are not necessary. There are rare cases but you know what - they do not fit my 80/20 rule. They are the glaring exceptions, not the norm.
Don didn't share any real world experiences -- do you see a case study there? I didn't?
I write everyday about how to use Oracle, every single day.
I write everyday about how Oracle actually works. Every. Single. Day.
Is this a war?
Bob B, September 14, 2005 - 3:10 pm UTC
This thread seems like a war. A lot of people are getting shot up needlessly and in the end, no one will be the better for it. Agree to violently disagree and let this thread go back to being technical. There are like 100 comments on this thread and about 3 of them have anything to do with Oracle or the way it works - WTF?
What is your experience with PGA supersizing?
A reader, September 14, 2005 - 3:26 pm UTC
Don’s tip was aimed at DBA’s, and I’ve heard you say that you are not a DBA. I am a DBA, a real DBA with years of real experience. I understood exactly what the point of the tip was and your suggestion that it be removed was, in your own words, “wrong”. This attack was completely unwarranted and very unprofessional.
I wasn’t born yesterday, and thankfully, neither was SearchOracle. You talked about the removal of the tip in several of your blogs, and now you bring it here. What’s with that Tom,? Even a week later you started a whole new blog topic asking “I just wonder why the article still exists?”.
Tom, you posted - “there was nothing to debate here, this was not an opinion piece”
Why then, did you later call it a debate and why didnÂ’t you invite Burleson to join in?
I’m confused. You say “Everything in that article could have been proven (proven wrong in this case).” and Burleson says that it was substantially correct?
It canÂ’t be both, and IÂ’m a beginner and I donÂ’t know who to believe. Who is lying here?
And by the way I have 5 Rampant books and I have found them all to be excellent. I ask you, where else can you get a 200 page book for $9.95?
September 14, 2005 - 3:56 pm UTC
did you read don's tip, if so, did you find it "correct" and or "accurate"
if so, please see
</code>
http://www.jlcomp.demon.co.uk/untested.html <code>
there were so many things *wrong*.
unwarranted, I do not believe you read it then.
tell me, what was the crux of the tip? What was it's point.
And did it accomplish that?
Simple questions -- very simple.
I didn't call it a debate.
Why would I invite anyone to anything?
You have the information on Jonathan's site.
You have the information right above you.
Have you read it?
If dollars per pages is your buying criteria, so be it.
Bravo SearchOracle!
DonÂ’t smear Oracle, September 14, 2005 - 3:34 pm UTC
I smell a rat here. I also liked DonÂ’s tip, and I like it better when it was short and sweet, before the train wreck started. DonÂ’s explanations were clear and unambiguous, he made his point, and nobody other than you felt it necessary to pick at it.
I followed your blog Tom, with the fascination of someone watching a train wreck happen and I left feeling sick to my stomach. Tom, you didnÂ’t seem interested teaching or clarifying, it was just you sticking up for JonathanÂ’s crappy proof and that that you were somehow justified in requesting that SearchOracle remove the tip. DonÂ’s rating was 4.3 out of 5 before you trashed it, and SearchOracle has some very knowledgeable readers, far more savvy than those I see here.
Tom, Oracle Corporation has invested a fortune into making you a household name in the Oracle world with full technical control over OTN and Oracle Magazine tips, your regular column on Oracle Magazine, and dozens of Oracle paid-for lectures and conferences.
Ten years ago Oracle experts were judged by their merit, not how much money their employers spend to promote them. We all knew Niemiec, Ensor and Ault, but IÂ’d never heard of you. Face it Tom, its Oracle investment in you that created your huge fan base. Otherwise you would have risen to the top by your own merit like the others did, 10 years ago.
What in the world makes you think that you have any right to ask a publisher to remove their work? I applaud SearchOracle more than ever, DonÂ’s restraint in not dignifying this series of unfounded technical attacks. Tom Kyte, IÂ’ve lost all respect for you. You remind me of the schoolyard bully.
Bravo SearchOracle, for standing up to the Tom Kyte the backyard bully. Tom, resist the temptation to remove with some lame reason like you did with Mike AultÂ’s post. It has legitimate technical issues and it has every much right to be here as the insults against Don Burleson and the Rampant authors.
September 14, 2005 - 4:01 pm UTC
short and sweet and wrong. cool.
I suggested to searchoracle, after providing them with 8 factual inaccuracies in the paper, that perhaps they might consider retracting it yes.
I clearly stated that.
Bravo for getting it almost fixed after two enhancements (seems some people are forgetting that - that the crappy work by Jonathan led to somewhat major revisions and 180 degree turns in statements. Amazing isn't it)
I'm flattered you believe I have so much control, but as stated above - I do not.
Once upon a time ago, I reviewed all user contributed tips (weeding out things that demonstrated how to update the data dictionary for example).
But I gave that up a while ago to someone else.
Laughing out loud at the powers you ascribe to me.
What is your experience with PGA supersizing? REALLY A DBA?
Bill S., September 14, 2005 - 3:36 pm UTC
So, if you REALLY are a DBA with years of 'real-world' experience, kindly show us all where the original article was completely right and where Tom or Jonathan's criticism of it is completely wrong. Because, my so-called DBA, THAT'S what it is all about, has been all about. Fire up your test database and provide us with some proof please.
Just a little. Please.
about test cases and proofs
A reader, September 14, 2005 - 3:40 pm UTC
By the way, can you comment your statement that “a single exception proves wrong”? Maybe that’s true in math but it’s not true for databases, right? Me, if something works on 95% of databases, I want to hear about it. It would be a lie to say that something that applies to 99% of Oracle databases is wrong because of a rare exception, right?
September 14, 2005 - 4:03 pm UTC
You don't know that it works on 95% of the databases. You haven't worked on 95% of the databases.
Me, I'd rather know the caveats. I'd rather know what I'm doing then just to try and throw darts, shoot guns into rooms, and see what happens.
And yes, when I did the proof above for the 10mb comment, that applied 100% of the time, not 5%.
So - well, don't know what else to say. That would be a test case that conclusively proves that the stated 10mb limit was *false*. It was not a limit.
To that supposed DBA
Bill S., September 14, 2005 - 3:51 pm UTC
By the way, just curious about this -
This quote of yours:
[quote]I am a DBA, a real DBA with years of real experience.[/quote]
kind of conflicts with THIS quote of yours:
[quote]It canÂ’t be both, and IÂ’m a beginner and I donÂ’t know who to believe. Who is lying here?[/quote]
Don't you think?
September 14, 2005 - 4:04 pm UTC
;)
A rat indeed
Jeff Hunter, September 14, 2005 - 3:53 pm UTC
[quote]I smell a rat here.[/quote]
Funny, I smell a troll. Either way, it stinks.
[quote]nobody other than you felt it necessary
to pick at it.[/quote]
Um, it was somebody else asking what his opinion was.
[quote]
Tom, you didnÂ’t seem interested teaching or clarifying
[/quote]
I can't even begin to imagine how this could be half-true.
[quote]
Tom, Oracle Corporation...with full technical control over OTN and Oracle Magazine tips, your regular column on Oracle Magazine[/quote]
Tom are congratulations in order? You've been promoted?
[quote]
...but IÂ’d never heard of you.
[/quote]
are you serious? You know Tom has a column in Oracle Mag, he has a blog, you post on asktom.oracle.com, and you've never heard of him.
[quote]Face it Tom, its Oracle investment in you that created your huge fan base. Otherwise you would have risen to the top by your own merit
like the others did, 10 years ago.
[/quote]
Um, "a reader", I have known Tom for 10+ years. I first became aquaited with Tom when he answered one of my questions on comp.databases.oracle.server. Tom has risen through the ranks of Oracle because he helps people. Tom's reputation is stellar.
[quote]
What in the world makes you think that you have any right to ask a publisher to remove their work?
[/quote]
Because it's wrong and they have a professional oblication (or at least should have) to be responsible for the content on their site.
[quote]
Tom Kyte the backyard bully.
[/quote]
From "ambassador of oracle" to "backyard bully", how quickly he has fallen. I believe it is the schoolyard bully who sends his henchmen in to do his dirty work, eh?
David Aldridge http://oraclesponge.blogspot.com, September 14, 2005 - 3:55 pm UTC
Commenters are not just strongly divided on their opinions here, but also on their use of different characters for double-quotes.
Among those who say that the criticism of Don's article was unwarranted and unprofessional, the “” style predominates, wheras among others the more conventional "" style rules.
Just an observation :D
September 14, 2005 - 4:08 pm UTC
doh, now the 'tell' has been told
Supposed DBA
A reader, September 14, 2005 - 4:00 pm UTC
He begins saying that he is a "real DBA, with years of experience"
Then towards the end says:
"I’m confused. You say “Everything in that article could have been proven (proven wrong in this case).” and Burleson says that it was substantially correct?
It canÂ’t be both, and IÂ’m a beginner and I donÂ’t know who to believe. Who is lying here? "
...Its soooo funny.. He does not even know what he is talking about.
Hello, here is the difference: Tom "proves" his point with examples/test cases which you can reproduce yourself. Don, on the other hand smoke something funny.
95% of what?
Jim, September 14, 2005 - 4:10 pm UTC
If something works 95% of the time in a database then clearly the parameters or fence around the problem are not well enough known or defined.
In your question you have tried to suggest a link that doesn't exist. There was never a claim that under some circumstances this works 95% (or a similar type of %'tage; I am not hung up on 95%) of the time. There was a claim that this works always. (100% of the time) In that case some investigation by others (Tom, Jonathan) demonstrated that the 100% claim was false. Not that it was false 5% of the time, but false 100% of the time. They published enough factual information that others could check their work. So Don could have looked at the work they did and confirm that what they found was true or not. If not then he could have published his findings including how he got to the conclusion.
Instead Janet et al. chose to address these technical facts as if it were some political campaign for president. Gross. Why the Classics Professor posted is beyond me. He has every right to read and post, but I find it hard to believe that he "stumbled upon" Tom's site. More likely, and I have no proof, Don or his cohorts are taking this whole thing over the top and have sent out some missive about this and have asked for people to respond. Clearly these "Defenders of Don" don't really feel that good about it because most of the postings are anon.
Don's Defenders, take a breath and have someone you know and respect that isn't in the middle of this read what you wrote (without kibbitz) and ask them for their opinion.
Yes, I've removed some tidbits here and there
Thomas Kyte, September 14, 2005 - 4:13 pm UTC
That had nothing to do with the thread.
to the one poster:
I don't care if someone has referred to themselves using a certain term. It was just too inflamatory here, didn't serve any good purpose. Please just bear with me on that one.
to the other (Jurij), about certain books:
sort of the same thing, just didn't need it and the hassles that come with it. Not passing judgement as to the accuracy or whatever of it, just don't need it right now.
If we take the high road, that leaves the other road.
- Tom
Hmmm...
Bob B, September 14, 2005 - 4:14 pm UTC
<QUOTE>
By the way, can you comment your statement that “a single exception proves wrong”? Maybe that’s true in math but it’s not true for databases, right? Me, if something works on 95% of databases, I want to hear about it. It would be a lie to say that something that applies to 99% of Oracle databases is wrong because of a rare exception, right?
</QUOTE>
No one has shown that the tip works on ANY database. So far, the tip works on 0% of the databases and has been shown to NOT work on 100% of the databases. In my book, that's a BAD tip.
As for the exception to the rule comment, I agree. If something works on most databases, I want to know about that something. I also want to know under what conditions it might not work and how to detect that.
10 years ago
Ram, September 14, 2005 - 4:26 pm UTC
"We all knew Niemiec, Ensor and Ault, but
IÂ’d never heard of you. Face it Tom, its Oracle investment in you that created your huge fan base. Otherwise you would have risen to the top by your own merit like the others did, 10 years ago. "
Ten years is a long time, friend. Ten years ago I was going to school. Few of my classmates I was playing and growing up with 10 years ago are doctors now and I would go to them for treatment now if I have ailment, which I would not have done 10 years ago. 10 years ago my uncle was working in railways fitting things properly, he wouldnt be hired today because he is too old for that job.
10 years ago Ault, Ensor, Niemic might have been Oracle greats (I would not know). They need not be today. 10 years before Tom might not be as well known (Again I dont know). So what? But today he is well known because he happens to prove his points. Who knows where we will all be in 10 years from now.
Please stop calling Tom a backyard bully. I do not like it.
On my comment being removed
Jurij, September 14, 2005 - 4:31 pm UTC
<quote from Tom Kyte>
to the other (Jurij), about certain books:
sort of the same thing, just didn't need it and the hassles that come with it.
Not passing judgement as to the accuracy or whatever of it, just don't need it
right now.
</quote>
I totaly understand your position and - for the record - I don't mind you removed my post. Nonetheless, just FYI, what I have written about certain books from certain author is 100% truth and is 100% proovable.
September 14, 2005 - 4:51 pm UTC
Peter K, I took one of yours as well.
I do not think it fair to the good authors out there to have wide ranging statements made like that.
I'm sorry to do that, just, well.... Let's take the higher road.
DateÂ’s Incoherence Principle
Gabe, September 14, 2005 - 4:35 pm UTC
They will go on and on and on Â… what other options do they have? Â… Logic? Integrity? What needed to be said has been said, conclusion [technical or otherwise] have been drawn. LetÂ’s not give them the satisfaction of seeing Tom chopping/closing the thread.
Let them dribble their nonsense till the cows or horses come home Â… and you please abstain.
Please abstain. That could be a powerful message too.
=====================================================<<
David Aldridge http://oraclesponge.blogspot.com, September 14, 2005 - 4:40 pm UTC
>> doh, now the 'tell' has been told
Heh heh heh, sorry -- I thought it was pretty obvious though, and it's a shame to keep something like that to yourself :D
who you know and what you know
Niall Litchfield, September 14, 2005 - 4:43 pm UTC
Stephen Cerutti wrote
"I am a Rampant Tech author. I am not an expert on Oracle --I don't even know
what Oracle really is. I am a Professor of Classics with a Ph.D. from Duke
University and I teach courses on Greek and Roman antiquity. I know the entire
Burleson family, have had both of Don's children in my classes, and am proud to
be the first author of Rampant Tech's new "Pedagogue Series" that is going to
expand Don's press into the academic world. What does this say about Don? That
he doesn't care about education? And what is education, if not about
disseminating knowledge. I have read these reviews and find them to be for the
most part personally motivated. I have not read the "article" in question. I
probably wouldn't understand a word of it."
I certainly have at least *some* personal motivation in where I come from, I'm sure many of us do. I care about getting things right. That said it really is a bit unfortunate that you don't understand the articles technical merits. It does hamper your suggestion of personal vendetta.
Don said in effect.
If you want to use more resource *than Oracle would otherwise use* then set these unsupported parameters (and this supported one). That was the tip. It also included the comment that people who set a specific combination of parameters didn't understand what they were doing.
There were two big issues with the tip.
First if you did set the unsupported parameters (thus losing support you pay thousands of dollars a year for) then you would use *less* resource than if you hadn't touched them.
Second there is a whole class of users who knew exactly when and where parameters needed to remain set in combination - that is those whom Don said 'don't understand memory management' might well have understood it better than the author.
You asked 'what is education if it is not about disseminating knowledge?' The base problem with the article is that the knowledge disseminated is of the same calibre as that that says the world is flat. It isn't disseminating knowledge, its disseminating error. This isn't personal opinion either - its measurable and demonstrable.
If I published an article stating that Plato was Roman and held that the world was not in any sense real, but a construct of experience, then I'd hope you or your colleagues would suggest that I didn't know what I was talking about. I'd also hope that no-one defended me on the grounds that I really cared about education and so what I'd written must be true.
Lets get technical
A reader, September 14, 2005 - 4:47 pm UTC
OK, SERIES OF QUESTIONS TOM, PLEASE ANSWER EACH IN KIND:
LEWIS IS ACCUSED OF DECEPTION BY MIKE FOR NOT MENTIONING THAT HIS PROOF WAS USING SHARED SERVERS, A FEATURE THAT IÂ’VE NEVER SEEN USED IN A REAL DATABASE. DON SAYS ITÂ’S RARE, VERY RARE AND HE QUOTES YOU AS SUCH.
</code>
http://www.dba-oracle.com/t_mts_multithreaded_servers_shared.htm <code>
HE NOTES YOUR RULES OF THUMB! HA HA, HE HAS A SENSE OF HUMOR.
QUESTION 1: HOW MANY WEEKS EACH YEAR DO YOU SPEND ON CLIENT SITES?
QUESTION 2: WHAT PERCENTAGE OF SITES USED SHARED SERVER IN YOUR OBSERVATIONS
QUESTION 3: YOU SAY THAT EVERYTHING IN DONÂ’S ARTICLE CAN BE PROVEN WRONG WHILE JONATHAN SAYS PARTS ARE SUBSTANTIALLY CORRECT. WHO IS CORRECT, PLEASE? IT CANNOT BE BOTH.
QUESTION 4: DID YOU GIVE DON AN OPPORTUNITY TO TALK ABOUT IT BEFORE STARTING YOUR PUBLIC BLOGS?
QUESTION 5: WHY DID YOU DELETE MIKE AULTÂ’S POST AND YET LEAVE YOUR COMMENT ABOUT IT? DO YOU ALLOW MIKE AND DON TO POST HERE?
QUESTION 5: EVEN IF THE TIPS WAS WRONG, WHY NOT REQUEST THAT DON IMPROVE IT? WHY DID YOU ASK FOR IT TO BE REMOVED?
QUESTION 6: I WANT TO KNOW TIM DICHIARAÂ’S OBSERVATIONS. WILL YOU ASK HIM TO JOIN US?
I KNOW THAT THESE ARE NOT YOUR ONLY ISSUES, BUT LET US START THERE. THANK YOU.
September 14, 2005 - 5:01 pm UTC
If it were rare, why do we have bugs filed against it?
Many of our databases run shared server, search asktom, you'll see lots of people using it. It comes up all of the time.
(is your capslock stuck?)
He didn't quote me saying that, my advice was "be sure you actually need it, it solves a specific problem, if you don't need it don't use it"
Where do you see me saying "no one uses it in real life"?
My rule of thumb if you will is
"if you don't have a need for feature X, don't use it"
so what?
1) none of your business frankly. Just got done with one however, so - at least one. I'll ask you though, it is a funny request, what difference would it make in assessing the correctness of this particular article? I'm curious??
2) I've never kept statistics, but based on bug reports, based on questions observed on the newsgroups - more than one.
By the way -- does it matter? Was the statement
a) correct
b) incorrect
??? which one, pick one please.
3) did you actually read jonathan's paper? I don't *think so* :) Either that or tongue in cheek humor is lost on you.
4) I've tried that route in the past. It is not pretty.
5) Because Mike Ault has sent letters and hosted web pages complaining about this site and people impersontating him. I could not verify that the ip address was in fact Mikes - therefore, I felt removing it was the best course of action.
What would you have done? I removed it because Mike has expressly stated in writing to Oracle that he is upset that people impersonate him here, so I removed the ability for that complaint to be made in the future.
5 (again 5) Because, I've tried that in the past. I get challenged to technical pistols at dawn, I get other things said. I don't need it.
6) you can, searchoracle has lots of links to that effect.
PLEASE LET JUNJI SPEAK
A reader, September 14, 2005 - 4:52 pm UTC
HE SAYS HE CAN PROVE IT AND THAT IS GOOD ENOUGH FOR ME.
PLEASE ALLOW HIM TO SPEAK HIS TRUTH.
PLEASE DO NOT EDIT YOUR FORUM.
Alberto Dell'Era, September 14, 2005 - 4:56 pm UTC
> 10 years before Tom might not be as well known
> (Again I dont know).
10 years ago I was following Thomas Kyte and Jonathan Lewis on comp.databases.oracle.server - believe me, you would have noticed them immediately :)
In fact I built my own askTom and askJonathan by adding to my search queries on deja.com:
(thomas AND kyte) OR (jonathan AND lewis)
Good old days ...
Some Answers
Warren, September 14, 2005 - 4:57 pm UTC
Here are some answers (not from Tom) to your questions. By the way, no need to SHOUT!
QUESTION 1: HOW MANY WEEKS EACH YEAR DO YOU SPEND ON CLIENT SITES?
Answer 1: I spend about 40 weeks a year at clients sites as an Oracle DBA consultant.
QUESTION 2: WHAT PERCENTAGE OF SITES USED SHARED SERVER IN YOUR OBSERVATIONS
Answer 2: I see approximately 25% of my clients using shared servers.
QUESTION 3: YOU SAY THAT EVERYTHING IN DONÂ’S ARTICLE CAN BE PROVEN WRONG WHILE JONATHAN SAYS PARTS ARE SUBSTANTIALLY CORRECT. WHO IS CORRECT, PLEASE? IT CANNOT BE BOTH.
Answer 3: Probably worth your time to actually read Jonathan's articles and not the BC employee mis-quotes. The substantially correct comment was tongue-in-cheek and actually meant substantially incorrect (in my opinion).
QUESTION 4: DID YOU GIVE DON AN OPPORTUNITY TO TALK ABOUT IT BEFORE STARTING YOUR PUBLIC BLOGS?
Answer 4: can't answer this one
QUESTION 5: WHY DID YOU DELETE MIKE AULTÂ’S POST AND YET LEAVE YOUR COMMENT ABOUT IT? DO YOU ALLOW MIKE AND DON TO POST HERE?
Answer 5: ditto
QUESTION 5: EVEN IF THE TIPS WAS WRONG, WHY NOT REQUEST THAT DON IMPROVE IT? WHY DID YOU ASK FOR IT TO BE REMOVED?
Answer 6. Because the tip was wrong and dangerous. Tom is trying to protect the unwary out there. People who might implement an incorrect tip without testing first (as the author did).
QUESTION 6: I WANT TO KNOW TIM DICHIARAÂ’S OBSERVATIONS. WILL YOU ASK HIM TO JOIN US?
Answer 7: not sure. Why don't you ask him to join us!
Please stop with the Caps
Rich, September 14, 2005 - 4:59 pm UTC
"A reader",
Please stop using CAPS, it is quite annoying.
I do have a question for you though:
Why don't you identify yourself?
Rich
Caps Lock key on the left of the keyboard is not for keeping your hat on.
A reader, September 14, 2005 - 5:01 pm UTC
QUESTION 3: YOU SAY THAT EVERYTHING IN DONÂ’S ARTICLE CAN BE PROVEN WRONG WHILE
JONATHAN SAYS PARTS ARE SUBSTANTIALLY CORRECT. WHO IS CORRECT, PLEASE? IT
CANNOT BE BOTH.
Here is Jonathan's full quote
"I do agree, however, that some of the observations in the original article were “substantially correct” I would go so far as to say “absolutely correct”– specifically: there are some undocumented parameters that few people understand, and changing the _pga_maxÂ_size, _smm_max_size, and _smm_px_max_size can affect the memory available for workarea operations. The devil (as the saying goes) is in the detail."
I think there is a little humour there you appear to have missed. So apart from stating that there are undocumented parameters that can affect memory, everything else in the article is wrong.
So it appears without your selective quoting that both Jonathan and Tom can be, and indeed are, correct.
Re: PLEASE LET JUNJI SPEAK
Jurij, September 14, 2005 - 5:07 pm UTC
Mr "A reader":
Who in this thread is "JUNJI"? Whom are you refering to??? I belive you've done simmilar thing before - so, is it becoming a habit?
And why do you need to SHOUT at us? We read you loud (and no to clear) without SHOUTING!
Re Comment from Gabe "Please abstain. That could be a powerful message too"
pjm, September 14, 2005 - 5:15 pm UTC
I agree with Gabe, although by now, his followup is probably about fifty feet up the page :-)
I read the forum at lunch-time pretty much every day and in the evening, and I just hate seeing this thread getting bumped up to the top all the time.
I'm not a DBA or anything, just a humble writer of queries for MI purposes, and I've learned so much from this site about analytics, faking non-existent rows by selecting from dual, using group by (cube ...), all the really clever stuff that you clever guys know backwards. And I just sit there and puzzle it out for myself by searching the AskTom site.
Keep up the great work Tom, and so must all your regular contributors that make the site so great and informative: Gabe, Connor, Alberto, and even Mikito!!! (And so many others whose names do not spring to mind just now.)
That's enough waffle from me, and hopefully everyone else in this thread - GET BACK TO WORK EVERYONE!!!
To Cancunkid
Thomas Kyte, September 14, 2005 - 6:02 pm UTC
Cancunkid/or any other alias you might have used..
Someone posted as Mike Ault using the IP address you just used at 2:53pm east coast time today.
I've removed that post allegedly from Mike Ault, I have to continue doing that. Mike has been very clear in his desire not to be impersonated, I am just trying to help there.
So Cancunkid, if it was you, please don't do that, I'll just keep removing them. Thanks!
How odd...
Howard J. Rogers, September 14, 2005 - 6:17 pm UTC
And of course Mike's employer's own guidelines state that he must not post here, and it is therefore a pretty fair bet that it wasn't him.
Interesting that it is the (non-Mike) supporters of Don, that fearsome foe of anonymity and staunch advocate of the dangers of impersonation on the Internet, that nevertheless seem to resort to, er, anonymity and impersonation here. Odd that.
Just a Thought
A reader, September 14, 2005 - 9:37 pm UTC
I was reading 'Feynman Lectures on Physics' Volume -I. On the very first chapter on the very first page Feynman says
<Quote>
The principle of science, the definition, almost, is the following: The test of all knowledge is experiment. Experiment is the sole judge of scientific truth.
<unquote>
If only the 'World's renowned Oracle experts' can learn from a mere unknown scientist called Feynman... oh I am sorry ... don't want to start another fight about who is a scientist...As Bill repeatedly pointed out they always dodge their main mistakes and try to zig zag...
You know what I find odd, Howard?
Bill S., September 14, 2005 - 9:49 pm UTC
After all the "anonymous" postings here and such, they STILL haven't answered Jonathan's questions and they STILL haven't answered Tom's. I can live with not getting answers to my own, I'm just a lowly dba (tongue is firmly planted in cheek here so if you were sitting next to me, I'd sound loaded) but Jonathan and Tom have earned their spurs.
So, instead of continuing to post tripe and innuendo folks, why not try answering the questions? It's Double Jeopardy, where the score can really change in a hurry.
You know something, given past performances by these pro-article posters, I guess I don't find it odd at all - it's par for the course.
Contradictions galore!!
Dilip Patel, September 14, 2005 - 10:16 pm UTC
I am an expert DBA.
--- I am just begineer
Tom does not moderate his forums
--- Do not moderate your forum, Tom
:-)
Tom, you and Jonathan have given enough explanation/proof against original article. We are still waiting for answers to many questions raised in this thread to many anonymous readers. Until they answer, think it is best to ignore them.
Please advise Tom
A reader, September 15, 2005 - 12:07 pm UTC
I am a great fan of your work and I need to understand how much Oracle protects innocent people from being damaged.
I don;t feel comfortable with all of this name calling and unprofessional comments and before I enter this discussion I need to know if:
Does Oracle protect the rights of non-participants not to be unfairly defamed?
Does Oracle respect the laws that require publishers not to present someone in a false light in the public eye?
I have seen you remove so many comments without giving a full reason that I am confused about Oracle policies.
Also I am confused about why Mike Ault is not allowed to respond to the charges against him here. Is this an Oracle policy? Did you e-mail Mike to confirm his identity before removing his comments?
Please let me know. Thanks To You.
September 15, 2005 - 1:02 pm UTC
(so many questions, and you didn't answer mine from yesterday :( )
Please point out someone being portrayed in a false light. Please -- just point to it, call it out. Don't infer that "it must be there because I said it is there", call it out.
What "so many comments"? There were very few here that were actually removed.
And why would the removal of comments confuse it? Can you explain your reasoning there?
Obviously as you see - as anyone can see - it started to get personal. Things said about others - that I felt were beyond that which needed to be stated, I removed. Simple enough. They were about other individuals.
Things said about me - all 100% here. When I removed something, I made mention of it. They just didn't need be said.
As for Mike Ault, haven't you read many times on this page that
a) Mike Ault has officially complained to Oracle Corporate about being impersonated on this site.
b) I cannot verify it is Mike Ault actually posting, heck, the same IP address as used by Cancunkid is posting as Mike Ault? Sorry - that is too "strange" to ignore, don't you agree?
c) Mike Ault understands how he may contact Oracle corporation, he has done it before.
If Mike would like to post his comments, he can do so under the cloak of "anonymous" - if he makes a valid, sound argument - it matters not if his name is attached to it or not, does it? I mean, a sound technical, accurate argument with some evidence can come from anyone and would be treated the same.
Besides, B.C. has a clear policy forbidding their consultants from posting on this forum, perhaps that is why Harry stopped?
</code>
http://dba-oracle.com/consultant_forum_guidelines.htm <code>
This was posted on Orablogs this morning - seems releveant here.
Bill S., September 15, 2005 - 12:28 pm UTC
Folks, I found this this morning while perusing Orablogs, and thought it might make interesting reading. It did indeed.
</code>
http://laurentschneider.blogspot.com/2005/09/undocumented-parameter.html <code>
The blog entry above has this particular quote at the end (but you should read the entire entry, it is quite interesting):
[quote Laurent Schneider Blog entry]
Well, I must say that I finally opted for a more maintenable solution :
no more hints, no more undocumented parameter, but parallel processing up to 16 threads on a 4 cpus server.
As discussed in the iTar, a supported way to increased the maximum pga memory per single sql query is to increase the degree of parallelism.
As a rule of dumb, if you can avoid hidden parameters, avoid them![/quote]
Interesting, Mr. Schneider apparently did encounter some trouble with that undocumented parameter tip, so he opted to go with a more maintainable route. Interesting....
September 15, 2005 - 2:19 pm UTC
A funny thing about that example were the settings used (thanks for Laurent for following up, appreciate that).
But say we do this:
ops$tkyte@ORA9IR2> alter system set "_pga_max_size"=2000000000;
System altered.
ops$tkyte@ORA9IR2> alter system set pga_aggregate_target=6G;
System altered.
ops$tkyte@ORA9IR2>
ops$tkyte@ORA9IR2> select *
2 from all_parameters
3 where name = '_smm_max_size';
NAME
------------------------------
VALUE
-------------------------------------------------------------------------------
_smm_max_size
314572
<b>the target workarea size for that session is some 300meg, the 2gig setting on _pga_max_size was a tad "high"...
If the goal was to permit 2gig workareas in a serial batch processes, well, I might use this:</b>
ops$tkyte@ORA9IR2> alter session set workarea_size_policy=manual;
Session altered.
<b>and then</b>
ops$tkyte@ORA9IR2> alter session set sort_area_size = 1073741820;
Session altered.
<b>or</b>
ops$tkyte@ORA9IR2> alter session set hash_area_size = 2147483647;
Session altered.
ops$tkyte@ORA9IR2> alter session set sort_area_size = 2147483647;
Session altered.
<b>that is, for a serial batch job, we would not need to go this undocumented route at all unless you wanted to get to 4gig workareas </b>
But again, appreciate the followup by Laurent, that was great.
Maurice Mueller, September 15, 2005 - 12:59 pm UTC
Bill,
A year ago I made some intensive testing on 9.2.0.5 (64 bit) with _pga_max_size and I got the same ora-4030 as Laurent Schneider got. However after applying patch 3130972 (or upgrading to 9.2.0.6) the problem got fixed.
Regards
Maurice
Maurice - more details please?
Bill S., September 15, 2005 - 2:41 pm UTC
Reply to Maurice:
Were you doing this to solve a problem or just playing? If solving a problem, what problem? Did you contact Oracle Support before setting these undocumented parameters?
Apparently, Mr. Schneider seems to have found a supported, easy way to achieve his goal that didn't require fiddling with undocumented stuff. And (see Tom's reply above) he didn't even NEED to touch these parameters to achieve what he needed to in the first place! Mr. Schneider now apparently seems to think that simply playing with undocumented parameters is [quote]a rule of dumb[/quote]. My point was simply to provide someone else's take on the situation - someone who actually did this, why they did it, and what they finally concluded.
You also seem unsure as to what "fixed" the problem you had - are you sure it was truly fixed? And that your undocumented parameter setting is still set as you originally thought?
September 15, 2005 - 2:46 pm UTC
and don't get me wrong - is there a time and place to muck with these?
Sure, absolutely.
But - big big but - you better really understand how they interact, what they do, how they work and have a good understanding of how pga memory management itself works.
That is what was missing from the paper in the first place. The advice was backwards, the definitions wrong. That was the "danger". Misinformation coupled with backwards advice.
Rationale
Jonathan Lewis, September 15, 2005 - 3:00 pm UTC
I think the most interesting question in this thread is the one from VA:
What is the rationale behind this design decision?
And when you've got the answer to that question, you've also got the answer to Bill's question, and the generic question - when might you mess with hidden parameters.
Whenever I start to investigate a new feature I ALWAYS ask myself what the rationale is, whose problem it is solving, who Oracle Corp. designed it for.
In this particular case - who is it for ? Well, if one person is allowed 5% of the memory, that means you've used up all the memory if you get 20 people working (or 3.33 people if they love parallel queries). It's desgined for concurrent systems. The target is to keep lots of people happy, whilst allowing a few to get busy and still be happy.
Once you realise that, you can say things like: "But I've only got 5 concurrent users EVER on this massively huge system - the strategy wasn't designed for me - maybe I should change it."
You DON'T say - "this query doesn't go fast enough, maybe I should fiddle with these parameters."
Prediction for a near-future release:
The various PGA targets will become semi-automatic, dictated by a rolling average number of users in the recent past. The system will try to detect if it is an OLTP where concurrency has to be protected, or a DSS/OLTP mix or a full DW where a large fraction of the PGA target can be reserved for a small fraction of the users.
For Tom on False Light
A reader, September 15, 2005 - 3:47 pm UTC
Tom,
You want evidence of presenting smeone in a false lgiht? Here you are:
Looks like Janet Burleson is doing the same thing with 'Guide Horses'. Howard
Rogers has some excerpts on his blog about her lack of knowledge while acting
like an expert from someone completely outside of Oracle.
</code>
http://www.dizwell.com/libelous_link_remved" <code>
This link directs your readers to something that says that Janet Burleson placed peoples lives at risk without proper training. It's a lie, a hideous lie, and if you had bothered to check it out before allowing it here, you would know that.
asktom is a publisher, and publishers are required by law to check their sources. You cannot say that the anonymous posters are responsible for their own words, that's not how it works.
You already removed posts you judge offensive, how about this? Either do your job as a publisher or shut this forum down.
You erased two posts from someone claiming to be Mike Ault never bothering to e-mail him for verification.
I'm writing to Oracle about this embarassing web site, and I encourage others to do-so as well:
Oracle Legal
500 Oracle Parkway
Redwood Shores, CA
94065
How dare you allow people to post links to hurt people here? Shame on you, Mr. Kyte. I want to see you banned from participating in forums, including this one.
Doubt And Verify
Joel Garry, September 15, 2005 - 3:55 pm UTC
Ten years ago I was also a fan of Tom. And Mike. Tom for the obvious things he has become famous for, Mike for his V7 cookbook. I believe simple and straightforward DBA work _should be_ cookbookable, as Oracle has indeed gone towards. Once you get things going, then you can tune, tweak, develop, or whatever.
I like it when people try strange and mysterious things, it can help pull the curtains away. That only becomes bad when it is cookbooked. That is my major complaint with several of the Rampant books, and most of Don's tips. The information is not being processed or presented properly, and the target audience is incorrect. Worse, reasonable feedback is not being incorporated, and it appears the whole idea of academic review is corrupted.
Tom says Trust But Verify.
I'd add for these other investigations, Doubt And Verify.
For Tom on True Light
A reader, September 15, 2005 - 4:08 pm UTC
Tom,
Thanks for exposing Bogus Oracle experts.
Please give the address at Oracle where we can write to free you from mundane work so that you will have more time to spend on Ask Tom.
Please participte in more Oracle Forums and if possible can you please open more ways to share your tremendous knowledge on Oracle.
Alex, September 15, 2005 - 4:15 pm UTC
Sorry to clog up this tread some more Tom but I can't let that last post go without defending you:
To "a reader", that was absolutely pitiful, pathetic at best. Your threats of legal action are completely laughable. Thanks for posting the address though. Maybe I'll write Oracle telling them what a tremendous asset Tom is to the Oracle community. Thanks again.
"It's a lie ..."
Doug Burns, September 15, 2005 - 4:23 pm UTC
"A reader" (Hopefully you know which reader I'm replying to, ... there are so many)
"It's a lie, a hideous lie, and if you had bothered to check it out before allowing it here, you would know that."
Presumably you'll be able to post a link to a National Federation of the Blind retraction. I look forward to it.
"You erased two posts from someone claiming to be Mike Ault never bothering to e-mail him for verification."
So how do you know that? Did you speak to Mike, just to check?
"I'm writing to Oracle about this embarassing web site, and I encourage others to do-so as well:
Oracle Legal
500 Oracle Parkway
Redwood Shores, CA
94065"
Do I need an airmail stamp? Could be a little pricey, but worth it.
"Shame on you, Mr. Kyte. I want to see you banned from participating in forums, including this one."
Tee-hee ;-) I think we should ban Tom from *his* forum? Or maybe you meant CDOS? (I can't see those guys going for it ...) Maybe you meant the forums at dba-oracle.com? Isn't everyone banned from there already?
Sorry, Tom, I know this isn't very productive and I know we should all be above this but, really!
Doug
P.S. I accept responsibility for this post. I'm not anonymous.
To Doug Burns
A reader, September 15, 2005 - 4:40 pm UTC
You asked me to provide the NFB information, well that's not my job, it's Tom's jobs not to encourage people to read libel by allowing people to put links to it.
All he had to do was verify the source, you know, like every responsible publisher does. Gigi and Suzanne are in the phone book. Go ahead, call them, you will find out why republishing this lie was a very very nasty thing to do.
Also, talk to NFB member Ann Edie, a guide horse user. Read what she says about their resolution:
</code>
http://www.guidehorse.org/coop_panda.htm <code>
Oh, and check the public records. Maybe I'm hinting, you know, THE FEDERAL LAWSUIT Janet discussed in Helping Hooves? But hey it's all good, Tom will leave it there, he is too lazy to properly moderate this slug fest and don;t tell me that an author doesn't know the rules of verifying sources.
I'm sure we will be hearing more about this, in the TV and news media, probably. Oracle claims to be an ethical company, let's see them try to justify this.
Stop feeding the troll
Jim, September 15, 2005 - 4:52 pm UTC
People stop feeding the troll! Arguing with a fool only makes you look foolish.
Take this off-line?
Doug Burns, September 15, 2005 - 4:57 pm UTC
I soooo wanted to leave this - there are more useful and important things that could be discussed on this technical forum, but I admit I've been rather sucked in by this. Feel free to post a comment at </code>
http://oracledoug.blogspot.com/2005/09/ive-tried-not-to-but.html <code>and we could talk about it there and leave Tom and his readers to more interesting subjects
But, seeing as you raised this stuff :-
"Gigi and Suzanne are in the phone book. Go ahead, call them, you will find out why republishing this lie was a very very nasty thing to do."
International phone bills are expensive, perhaps you could explain?
"Also, talk to NFB member Ann Edie, a guide horse user. Read what she says about their resolution:"
Was that link meant to make me see differently?
"But hey it's all good, Tom will leave it there, he is too lazy to properly moderate this slug fest and don;t
tell me that an author doesn't know the rules of verifying sources."
Do you realise how ridiculous that makes you sound? Tom Kyte, that well-known slacker ....
"I'm sure we will be hearing more about this, in the TV and news media, probably. Oracle claims to be an ethical company, let's see them try to justify this."
Zzzzzzzzzzzzz .....
Stop feeding the troll
Doug Burns, September 15, 2005 - 4:58 pm UTC
Sorry Jim (and Tom, and the rest of you)
You're right. I'll stop now.
Still waiting
Jonathan Lewis, September 15, 2005 - 5:36 pm UTC
Harry Conway,
It's been about 48 hours since I asked:
"In passing - would you care to list a couple of the things in the original article which you thought were correct."
Since then, there's been a sad absence of things you thought were correct. Of course, absence of evidence does not constitute evidence of absence, but if you had the courage of your convictions, wouldn't you be able to post at least one or two tiny little points ?
Would you like some help to find them ?
1) The name at the top of the article was Don Burleson, that was probably correct.
On second thoughts, though, shouldn't that be Donald Keith Burleson ? He's even got his own name wrong ! But, hey, that "Donald Keith" option is just an obscure feature that hardly anyone uses, so the criticism is invalid. (That's irony, Harry, just in case you aren't familiar with the concept).
Yes, after due consideration of all the facts, the name of the author is almost certainly correct.
There are a couple of other equally trite points that you might also quote - but do try and find something that merits the title of 'leading expert' that Burleson likes to claim.
sarcasm attack
Alex, September 15, 2005 - 5:41 pm UTC
on publishing
Niall Litchfield, September 15, 2005 - 5:57 pm UTC
Yet another reader (I'd give my eye teeth for the reader's IP to be associated publicly with their comments ) wrote
"asktom is a publisher, and publishers are required by law to check their
sources. You cannot say that the anonymous posters are responsible for their
own words, that's not how it works."
You know this for a fact right? You know the relevant law that applies to asktom and how a court would judge it. perhaps you can quote the legal reasoning.
"You already removed posts you judge offensive, how about this? Either do your
job as a publisher or shut this forum down."
Is it only this forum, or other forums you have in mind as well. The dba-oracle forum has adopted a much more agressive editorial policy - banning people for disagreeing with the owners and so on, I do hope you will agree that that too is a forum that should be shut down.
"You erased two posts from someone claiming to be Mike Ault never bothering to
e-mail him for verification."
But Mike's employers ban him from posting here. He has complained about being impersonated here. Surely it would be daft to imagine that he would break his employers guidelines to post here.
"I'm writing to Oracle about this embarassing web site, and I encourage others to
do-so as well:
Oracle Legal
500 Oracle Parkway
Redwood Shores, CA
94065 "
embarassing in what sense, that it evidences behaviour and promotes the oracle database; that it offers evaluations of the answers given; that it allows users to evaluate their understanding. It surely couldn't be just that *in response to a genuine question about an article* it offered an explanation as to why the article was incorrect. No-one would write to Oracle legal because they, or someone they liked, had been shown to be incorrect. Would they?
To recap those questions.
1) what is the legal basis for your quasi-legal assertions, what is the appropriate jurisdiction for any putative case and what are your qualifications for so concluding.
2) do you agree that the oracle forum sponsored by burleson consulting should also be closed down? if not why not? is the legal issue any different.
3) why should anyone bother to mail Mike before removing comments here, given that he is not permitted to make any comments here by the express desire of his employer?
4) what exactly is embarrassing about this website that deserves the attention of Oracle legal. The promotion of Oracle products, the ability for Oracle users to enquire about the product for free, the evidence based nature of the site or what *exactly*? Remember how it started - an official Oracle publication publishing material that was bad. Is this where we want to go back to?
Hash
Dave, September 15, 2005 - 5:59 pm UTC
Can you put the MD5 hash of the ip address next to the name so we can see how many different names come from the same IP? :)
A reader's trolling - funny but way stupid
Dilip Patel, September 15, 2005 - 6:13 pm UTC
To "A reader",
Whoever you are, please give us a break. Take this trolling somewhere else. By accusing and threatening Tom and inciting others, you have proved how low you can get. You had lost techical respect in oracle comunity and now your writings such as above is taking you far even on personal front.
This is technical forum and just try to answer Tom/Jonathan queries. Stop wasting everyone's time and stop distracting everyone from real technical issue in this thread by talking about legal issues.
No regards,
Dilip
test
A reader, September 15, 2005 - 6:15 pm UTC
test
September 15, 2005 - 6:19 pm UTC
Note: due to the passion involved here which seems to have overcome the technical side, I've closed this thread to anymore followups.
Tom Kyte.
PGA
Reader, January 30, 2008 - 9:46 pm UTC
How I can limit the pga size to grow upto certain size, I am facing error 4030, when my oracle.exe process consume around 2gb of memory (Which is a limit for 32BIT), with PGA growing upto 1.4GB?
OR
Can we restrict pga size and redirect to temp?
February 04, 2008 - 2:52 pm UTC
PGA memory relies on the OS to swap (page to temp).
You would lower your pga_aggregate_target (or sort area/hash area size if you are using manual memory allocations) to decrease the pga size - but bear in mind, if a programmer codes:
declare
type array is table of long;
l_data array;
begin
for i in 1 .. 100000
loop
l_data(i) := rpad('*', 32765, '*' );
end loop;
end;
/there won't be a thing we can do about it (other then raise ora-4030)
If you configure shared server, you can move UGA memory from the PGA into the large_pool - the SGA. There you can set a resource profile to limit the amount of UGA memory they can allocate from the SGA
http://docs.oracle.com/docs/cd/B19306_01/server.102/b14200/statements_6010.htm#sthref7184
A reader, February 17, 2008 - 8:12 pm UTC
So can we say that the parameter pga_aggregate_target is similar to undo_retention in a way that this is just an advise to oracle and if required it will use more than the specified limit?
February 18, 2008 - 7:26 am UTC
yup
Can we refocus on the big picture?
Udi Karni, August 08, 2008 - 6:23 pm UTC
People are getting hung up on the hidden parms to try and resolve the bigger picture which is that Oracle uses PGA sparingly and "saves" it for additional queries.
If I have a server with 64GB of RAM that does batch data warehouse and I allocate 60GB to the PGA - Oracle will not exceed about half of that for query one, a little more for query two, etc.
How can we get Oracle to give 90%+ of the PGA to query one since a single query maxes out the server anyway? and not "save" half of the PGA for additional queries that we never submit because all they do is slow query one down.
Hardware vendors try to sell us all this RAM and Oracle never uses more than half of it. That's hard to justify, and very unfortunate because when using RAM Oracle blazes, and then slows down when it switches to disk based TEMP.
Any way to "explain" to Oracle that this server runs huge batch jobs one at a time and can it please allocate 100% of the machine to the job at hand?
Thanks.
August 12, 2008 - 4:25 am UTC
... If I have a server with 64GB of RAM that does batch data warehouse and I
allocate 60GB to the PGA - Oracle will not exceed about half of that for query
one, a little more for query two, etc.
...
actually, it is much less than "about half", more along the lines of 2.5 to 5%...
You would use manual memory management for that special batch process, it can alter it's SESSION and say "I'm going to tell you I'm the only game in town, use this sort area, use this hash area size..."
... Hardware vendors try to sell us all this RAM and Oracle never uses more than
half of it. ...
that statement in itself is so so so wrong - please do not ever say that, someone might believe you.
But, if you want a single session to use a big PGA, just set the workarea policy to manual for that session and set the sort/hash area_sizes...
Sorry Tom - not quite
Udi Karni, August 21, 2008 - 6:28 pm UTC
Tom,
The notion of going back from 10g's fast and efficient PGA based memory manager to 8i's clunky MANUAL mode is horrifying, but I gave it a try anyway.
It won't work for us. The SORT_AREA_SIZE parm is limited to 2G, so it's obviously per parallel process or similar - not a total number. So - if the size is too big or our parallel degree is too high - the sum total exceeds total memory and Oracle blows up, which never happens in AUTO mode. Since we have a variety of SQL - we can't customize each and every one - so this approach won't work. We won't miss it....
Then again - in AUTO mode - if you run a single SQL on a server - you will not get more than 50% of the PGA. A 2nd SQL will get maybe another 15%, a 3rd - another 10%, etc.
If the developer that wrote the PGA memory manager insists that a single parallel SQL can get more than half the PGA - I have a bet for him which he will lose. I've seen this hundreds of times.
So - yes - on a server running 1 statement at a time - half the RAM goes unused, to the consternation of our hardware vendor.... and with 1 statement taking up the entire IO bandwidth - there's no point in running a second - which will just take away from the first - so "1 at a time" mode is where we have to be.
How hard is it to modify the PGA memory manager to also have a "DSS" mode - where the first statement gets 80-90% of the RAM, and the rest, if there ever are any, split the remainder? Just modify the 50% governor a little bit, based on a parameter the user can set to tell Oracle this is a "one big job at a time" DSS machine.
?????????????????????????????
August 21, 2008 - 10:29 pm UTC
more than half. Not sure where you get the 50% from? What is your reference on that. It is much lower.
.. the sum total exceeds
total memory and Oracle blows up, which never happens in AUTO mode. ...
never say never, it can easily 'blow up', not all of the pga is tunable, there are tunable and untunable bits - it can and does exceed the aggregate target at times.
OPen or Closed
Richard, August 22, 2008 - 8:16 am UTC
Why did this thread ever get reopened?
PGA AUTO mode is actually great
Udi Karni, August 22, 2008 - 2:55 pm UTC
Tom,
Yes, it is possible for the PGA memory manager to exceed and run out of memory. However, as compared with 8i's manual mode that would run out of memory 100 times as often - I believe 10g's memory manager developers deserve a big attaboy for their conservative and prudent logic.
For DSS, however, it's too conservative. You might be right about the <5% allocation - but that's per thread. Here's what happens in the big picture.
A server with 64GB of RAM, a 10GB SGA, and 50GB PGA will use 12-13GB at "rest". That's 10 for the SGA and 2-3 for OS and Oracle kernel.
You run a single parallel query, and in TOTAL - you will never, never, ever, never see memory utilization exceed 38GB. That's the initial 13GB plus 50% of PGA (0.5 * 50) which is 25GB for a total of 38GB.
Regardless of the parallel setting or anything you do - you buy 64GB and Oracle uses no more than 25GB for the first SQL. Since 1 SQL is all the system can handle IO wise - 25GB gather dust.
This is empirical, Tom, but we've seen this hundreds of times.
Since we can easily use a 256GB PGA - we have to ask management to buy 512GB so that the half that Oracle is willing to use equals the 256GB we need.
Buying 1/4TB of RAM that won't get used does not make Oracle nor us look very good....
Help. I am sure your developer friends will have an idea.
Thanks,
Udi
5%?
Stephen, August 28, 2008 - 3:08 pm UTC
I am still curious as to where this 5% is coming from. My finding indicate the following in 10gR2:
* The _pga_max_size is the greater of a) 200M or b) 20% of the pga_aggregate_target
* The _smm_max_size is the lesser of a) 20% of the pga_aggregate_target or b) 50% of the _pga_max_size
* The _smm_px_max_size is 50% of the pga_aggregate_target; not dependent on the _pga_max_size
--------------------------------------------------
104857600 pga_aggregate_target
209715200 _pga_max_size
20971520 _smm_max_size
52428800 _smm_px_max_size
20971520 20% of pga_aggregate_target
104857600 50% of _pga_max_size
--------------------------------------------------
209715200 pga_aggregate_target
209715200 _pga_max_size
41943040 _smm_max_size
104857600 _smm_px_max_size
41943040 20% of pga_aggregate_target
41943040 50% of _pga_max_size
--------------------------------------------------
419430400 pga_aggregate_target
209715200 _pga_max_size
83886080 _smm_max_size
209715200 _smm_px_max_size
83886080 20% of pga_aggregate_target
104857600 50% of _pga_max_size
--------------------------------------------------
629145600 pga_aggregate_target
209715200 _pga_max_size
104857600 _smm_max_size
314572800 _smm_px_max_size
125829120 20% of pga_aggregate_target
104857600 50% of _pga_max_size
--------------------------------------------------
1073741824 pga_aggregate_target
214732800 _pga_max_size
107366400 _smm_max_size
536870912 _smm_px_max_size
214748364 20% of pga_aggregate_target
107366400 50% of _pga_max_size
--------------------------------------------------
4294967296 pga_aggregate_target
858992640 _pga_max_size
429496320 _smm_max_size
2147483648 _smm_px_max_size
858993459 20% of pga_aggregate_target
429496320 50% of _pga_max_size
--------------------------------------------------
If this is incorrect someone please correct it.
August 29, 2008 - 11:03 pm UTC
it depends on version
it depends on what is going on
and you do not get all of the memory AT ONCE. Your session is expected to have more than one query going.
and you see the word "max" - as is maximum, as in "hopefully the upper bound" right? that is for your SESSION, and your SESSION is anticipated to have many queries.
stop looking at "_" parameters. They change from dot to dot to dot to dot.
I've been talking about individual queries, your PGA is more than "a single query"