| Oracle FAQ | Your Portal to the Oracle Knowledge Grid | |
 |
 | |||
| Oracle FAQ | Your Portal to the Oracle Knowledge Grid | |
|
| |||
Home -> Community -> Mailing Lists -> Oracle-L -> RE: Your opinion please :) explanation of my point of view
Okay, I usually only post when I have a question for the Oracle Guru's
on the list. I realized real quick like that the collective knowledge of
the Oracle Brainiacs here didn't need any of my help. I seem to be too
busy running around catching all this free oracle manna dropping from
the sky and wondering just what these folks eat/drink to know all this
stuff (They seem to have a common beverage but this RTFM they mention ..
Is that like a low carb food or what?)
Now, since you are asking my opinion on a document for DBA's that don't
know much, hey I figure I can qualify :)
I think your attempt to write a paper and share knowledge is
commendable. It's more than I could do at this point.
I have looked it over and have provided the following comments (FWIW) 1. I don't think the title should be 101 Tuning tips as much as "101 things that effect performance". Tuning is a process for identifying and performance issues. (Cary's book recommended here). 2. The 'resolves 90% of tuning problems.." statement is a dangerous one for a new DBA to hear. I have a whole bookshelf full of Oracle reference materials and 100 stacks of oracle articles (yours included) and I'd be safe to say that 100% couldn't solve my tuning problems if I have poor application design and it's running on a customers server that houses the app, the database, acts a domain controller and they all share the same RAID5 and 1G of ram. The only way a book or article will help me here is if I hit someone with it enough to get them to change. (let me savor this thought for a moment.....ok ) I think you paper is a good start, but I think the newer DBA's would be better served by less of a checklist approach and a little more of 'things that effect optimizer performance', things that effect index performance', 'Initial database parameters that effect performance'. More of a cause and effect approach with examples. That I believe would be more beneficial.
Sorry for the short novella, and thanks for letting me express my opinion.
David Phillips
Oracle 8i OCP
BAARF member#30
-----Original Message-----
From: Juan Cachito Reyes Pacheco [mailto:jreyes_at_dazasoftware.com]
Sent: Thursday, February 26, 2004 2:33 PM
To: oracle-l_at_freelists.org
Subject: Re: Your opinion please :) explanation of my point of view
Yes, this resolves the 90% of tunning problems in standar database
edition.
The reason is simple, in STANDARD edition (no parallel execution, etc.)
, at
least here in Bolivia, you don't have lots and lots of users, maximum
20,
the biggest table I have has 3,000,000 of records.
So the tunning problems are not really serious in 9i.
Now I challenge you which tunning problem is not considered in that
paper,
and should be :=)
> Well, if your paper "solves 90% of tuning problems", than this list
can
> rest peacefully...
>
> Igor Neyman, OCP DBA
> ineyman_at_perceptron.com
>
>
>
> -----Original Message-----
> From: oracle-l-bounce_at_freelists.org
> [mailto:oracle-l-bounce_at_freelists.org] On Behalf Of Juan Cachito Reyes
> Pacheco
> Sent: Thursday, February 26, 2004 10:34 AM
> To: oracle-l_at_freelists.org
> Subject: Re: Your opinion please :) explanation of my point of view
>
> thanks to you, this will improve a lot, the point is that I don't have
> too
> much time.
>
> If you have something brief, for exapmle I got from Tom Kyte,
"function
> in
> c+ can be up to 2 times faster", to have an idea when to use c and
when
> not,
> or from tom kyte to "try to do in plsql, if you can't use java and if
> you
> can't use c++", this are the kind of guide and test takes looooong
time
> to
> find.
> Even when some of you could think this is a list of tips I got
randomly,
> it
> is not, in that paper there is all the tunning tips you need to solve
> 90% of
> tunning problems in standard edition, so don't think I take randomly
> some
> tips and put in a paper briefly and said "this is my paper"
> ----- Original Message -----
> From: "Freeman, Donald" <dofreeman_at_state.pa.us>
> To: <oracle-l_at_freelists.org>
> Sent: Thursday, February 26, 2004 11:24 AM
> Subject: RE: Your opinion please :) explanation of my point of view
>
>
> Let me chime in my 2 cents here. I printed and reviewed your document
> and
> noted a few
> things that I will discuss with my co-DBA. I liked the article because
> in a
> short
> document you summarized a number of things that were important in your
> experience. Some of
> them seem relevent to what I am doing so I will spend some time to
check
> them out. As
> far as you posting it, I personally try to pass on tips and tricks as
I
> learn them and I appreciate
> that behavior in others.
>
> As what we do is often a competitive business I have not often
> encountered
> people personally
> who freely give away valuable information which they struggled to
gain.
> The
> top-level professionals
> who answer questions here are doing 'pro-bono' work for which I am
> grateful.
> It's the same advice
> that people pay them for. In real life I'm beginning to think it's
> pretty
> remarkable when somebody
> says, "Come on over, I'll show you how it works." The lists are very
> valuable to me. I'm not in a
> position to travel to conferences, seminars, and attend workshops to
> improve
> my skills and don't
> work for an employer who values (or pays for) this kind of stuff.
>
> I think you are pretty brave for posting your 'baby.' I have been
> holding
> my breath because
> I thought it might provoke a lot of disagreement or outright
> disparagement
> of your effort. At
> this point I don't know enough to say if what you have posted is
> altogether
> technically accurate but
> I always check 'free' advice anyway.
>
> You are correct, I think, in the way some people approach books. I am
> struggling to understand
> some of the features in Oracle and have quite an extensive library of
> Oracle
> books. Most of them
> I haven't read cover to cover. I tend to read them a chapter, or even
a
> paragraph at a time depending
> on what my issue is. I download a lot of papers, "how-too's", tips and
> tricks.
>
> The value of the books on the market to me is to answer the bigger
> questions
> of why and when (or how much)
> instead of what and how. Its always helpful to me when somebody says,
> "On an
> OLTP system is should be X and
> on a DS system it should be Y." It gives me a point at which to start
> looking. The oracle books usually
> provide the exact syntax describing how a feature or function is
> implemented
> but don't answer the question,
> for me, "Is this an appropriate thing to do in a small datawarehouse
> environment..." I know, "It depends."
>
> So, thanks for posting!
>
> Don Freeman
> Oracle 9i OCP
>
>
>
> -----Original Message-----
> From: oracle-l-bounce_at_freelists.org
> [mailto:oracle-l-bounce_at_freelists.org]On Behalf Of Juan Cachito Reyes
> Pacheco
> Sent: Thursday, February 26, 2004 9:05 AM
> To: oracle-l_at_freelists.org
> Subject: Re: Your opinion please :) explanation of my point of view
>
>
> Thanks for your opinion.
>
> Now if you think "kill lazy and bad dbas", I can't do too much about
> this.
> I think every one likes his job, and a paper like this can wake up his
> interest in reading to a bad dba, and help to introduce a good dba who
> is
> starting to read.
>
> Yes this is true, the idea is to introduce to all this people to
> tunning.
> I think if every one would read the Oracle document as they should,
50%
> of
> the book would be useless. If you take randomly a good tunning book
and
> seek
> the same in Oracle tunning documentation, 90% is in the Oracle
> documentation, at least!!
>
> But even when you read and got an OCP, you need some guide about what
> you should check first, better and simple explanations, for example
> about
> CURSOR_SPACE_FOR_TIME there is not a clear explanation about when
to
> use
> them.
> You need a guide that introduce to tunning. to make you understand
what
> is
> important and what is the most important.
>
>
> ----- Original Message -----
> From: "Igor Neyman" <ineyman_at_perceptron.com>
> To: <oracle-l_at_freelists.org>
> Sent: Thursday, February 26, 2004 9:53 AM
> Subject: RE: Your opinion please :) explanation of my point of view
>
>
> > If "newbie DBA" doesn't like to read, he/she'd better be in
different
> > business...
> >
> > Igor Neyman, OCP DBA
> > ineyman_at_perceptron.com
> >
> >
> >
> > -----Original Message-----
> > From: oracle-l-bounce_at_freelists.org
> > [mailto:oracle-l-bounce_at_freelists.org] On Behalf Of Juan Cachito
Reyes
> > Pacheco
> > Sent: Thursday, February 26, 2004 7:48 AM
> > To: oracle-l_at_freelists.org
> > Subject: Re: Your opinion please :) explanation of my point of view
> >
> > Thanks for your opinion, tellme what do you think, about this.
> >
> > Point by point
> > 1) About the set of checklist, I agree, but the goal is to introduce
> to
> > tunning, in the market there are several excellent tunning books,
> > (including
> > the documentation where is almost all), but people don't read them.
> > I decided something short, a checklist explaining very very briefly
> what
> > you
> > can tune, the most important, the first steps, for newbies, now if
you
> > want
> > more information, you can go to documentation.
> > For example about statistics, I had two pages of a BRIEF
explanation,
> > but no
> > body is going to read all that, because instead of 8 pages it will
be
> 60
> > pages, it will be preferable to read the documentation, but several
> > questions you should had read some moment, shows people don't like
to
> > read,
> > people who never took the time to read neither concept manual, so I
> > decided
> > to explain the most important, briefly.
> >
> > For example, about statistics, the explanation is very short
> > 1.1 Statistics
> > The optimizer works based on statistics, if they are old, or
> inaccurate
> > you'
> > ll get a wrong execution plan.
> >
> > You must recalculate statistics every time an important change had
> > happened
> > in it:
> >
> > . Periodically, based on normal changes in a database
> >
> > . After importing a big amount of data
> >
> > . When distinct values in primary columns change
> >
> > . After creating indexes and table
> >
> > 1.1.1 Test database
> > If this is the first time you get statistics, you must remember some
> > database has fixed execution plans, or any other consideration can
> cause
> > a
> > statistics recalculation cause serious problems, as general rule do
it
> > first
> > in ha test database, before doing in the production database.
> >
> > 1.1.2 Recalculating statistics
> > Oracle recommends DON'T USE ANALYZE to gather statistics, the
package
> > USE
> > DBMS_STATS, this package get more statistics, specially for new
> > features.
> >
> > Don't execute DBMS_STATS on SYS schema.
> >
> > This command has dozens of options, like parallel execution, etc.
etc.
> > you
> > must read them.
> >
> > To gather all statistics
> >
> > EXEC DBMS_STATS.GATHER_DATABASE_STATS();
> >
> > To gather statistics in a schema:
> >
> > EXEC DBMS_UTILITY.ANALYZE_SCHEMA('ADM','COMPUTE');
> >
> > To gather statistics in a schema, more precisely, the one we use,
> > because
> > our database is a small database:
> >
> > EXEC DBMS_STATS.GATHER_SCHEMA_STATS( OWNNAME=>'ADM',
> > ESTIMATE_PERCENT=>100,METHOD_OPT=>'FOR ALL COLUMNS SIZE SKEWONLY');
> >
> >
> > What you understand after this is
> > -optimizer needs statistics
> > -changing statistics can cause you problems, first use a test
database
> > -use dbms_stats not analyze
> > That's all, and I think this is the basic thing you have to
> understand,
> > I
> > know dbas (not ocps) who don't know that simple fact," they have to
> > recalculate statistics periodically". Because their database are
small
> > and
> > comparing with sql server it takes a similar amount of time to
> process,
> > they
> > don't see it as a problem or a mistake.
> >
> > 2)About the optimizer_index_caching I read a lot, because we have a
> > problem,
> > we solved setting this parameter, about this parameter I say:
> > "This parameter sets the possibility to find blocks accessed through
> an
> > index in memory.
> >
> > Default value is 0, 0 means that no one block accessed through index
> > will be
> > found in memory THIS IS FALSE. A value of 90% is advisable"
> >
> > Could you giveme an example how could I explain it better.
> >
> > 3) About
> > "but pays no attention to individual business processes (or indeed
> > development modules)."
> > Could you giveme an example what for example I could add ,
> > But I'm not explaining in this paper things like ORDERED hint,
because
> > for a
> > newbie, first he have to do the basic things, once he had done all
of
> > them,
> > if he continue having problems, then the correct way is to read a
more
> > serious document.
> > Because if you give advanced points, he will try them first, is like
> > special
> > hints, I think you heard about them, they are not in documentation
> > because
> > it could cause problems to new dbas, because they could try to use
> them
> > to
> > solve problem, when the reality is they can be used only
> exceptionally.
> > but
> > there are book out there once you get the enough experience you
could
> > buy
> > them.
> >
> > Now if you want another point of view, or a better idea about this,
> you
> > are
> > welcome :).
> >
> > ----- Original Message -----
> > From: "Niall Litchfield" <n-litchfield_at_audit-commission.gov.uk>
> > To: <oracle-l_at_freelists.org>
> > Sent: Wednesday, February 25, 2004 7:29 PM
> > Subject: RE: Your opinion please :)
> >
> >
> > hi
> >
> > I have to admit I don't like the paper much. it looks like a set of
> > checklists with little or no reasoning behind them. for example the
> > suggestions on optimizer_index_caching have just start at x and work
> > from
> > there. where x is apparently an arbitrary number.
> >
> > my deeper problem is that the article seems to suggest that it is
> > possible
> > to tune systems but pays no attention to individual business
processes
> > (or
> > indeed development modules). I disagree, the logical unit for tuning
> is
> > the
> > thing that runs badly, the logical time to do it is when writing the
> > code
> > for it. setting db wide parameters is sticking plaster on the
titanic.
> > perhaps.
> >
> > Niall
> >
> > -----Original Message-----
> > From: AC.GWIA.oracle-l_at_freelists.org
> > [mailto:AC.GWIA.oracle-l_at_freelists.org]
> > Sent: Wed 25/02/2004 18:41
> > To: jreyes_at_dazasoftware.com; oracle-l_at_freelists.org
> > Cc:
> > Subject: Your opinion please :)
> >
> >
> >
> > Hi I took sometime to do a paper, to help newbies in tunning, this
is
> > not to
> > become they expert, but at least they know there are statistics in
the
> > database
> >
> > If you get some time, and give some opinion about it, like the
> previous
> > mistake I did with IN function, you are welcome.
> > Or something important I'm missing (for newbies dbas of course).
> > Did you remember the first time you had a tunning problem, before
the
> > ocp,
> > and you didn't knew where to start to check,and lost long time,
> changing
> > some database parameter to see if that improved something, the idea
is
> > to
> > save that step.
> >
> >
> >
>
http://www.geocities.com/juancarlosreyesp/101basictunningonOracle9i.pdf
> >
> > Juan Carlos Reyes Pacheco
> > OCP
> >
> >
> > ----------------------------------------------------------------
> > Please see the official ORACLE-L FAQ: http://www.orafaq.com
> > ----------------------------------------------------------------
> > To unsubscribe send email to: oracle-l-request_at_freelists.org
> > put 'unsubscribe' in the subject line.
> > --
> > Archives are at http://www.freelists.org/archives/oracle-l/
> > FAQ is at http://www.freelists.org/help/fom-serve/cache/1.html
> > -----------------------------------------------------------------
> >
> >
> >
> >
> >
-- Archives are at http://www.freelists.org/archives/oracle-l/ FAQ is at http://www.freelists.org/help/fom-serve/cache/1.html ----------------------------------------------------------------- ---------------------------------------------------------------- Please see the official ORACLE-L FAQ: http://www.orafaq.com ---------------------------------------------------------------- To unsubscribe send email to: oracle-l-request_at_freelists.org put 'unsubscribe' in the subject line. -- Archives are at http://www.freelists.org/archives/oracle-l/ FAQ is at http://www.freelists.org/help/fom-serve/cache/1.html -----------------------------------------------------------------Received on Fri Feb 27 2004 - 09:53:35 CST
 |
 |