Re: Which Doc to Read - 10g or 9i ?

From: HansF <Fuzzy.Greybeard_at_gmail.com>
Date: Fri, 30 Jun 2006 14:38:38 GMT
Message-Id: <pan.2006.06.30.14.40.20.50366@gmail.com>

On Fri, 30 Jun 2006 09:53:37 -0400, meathammer wrote:

> mccmx wrote:

>> I totally disagree with that comment.... There are some books out
>> there which contain information that you won't find anywhere in the
>> Oracle documentation....
>>
>> e.g.
>>
>> Cary Milsap's "Optimizing Oracle Performance"
>> Steve Adams' "Oracle 8i Internal Services for Waits...."
>> Jonathan Lewis' "CBO Fundamentals",
>> Dan Tow's "SQL Tuning", etc, etc
>>
>> Matt

> 
> But then the question is, are those extra info something the Oracle
> professional SHOULD know to do job competently, in general?

[Reader's digest summary: yes, these book have information, especially methods, that are not (or not easily) found in the documentation.]

The online documentation is fairly comprehensive. It has the general disadvantange of not 'flowing' so it is often difficult to get specific information unless you are familiar with it. It also presents a lot of facts, but does not always provide the context in which those facts are relevant. You may have to jump a lot between sections and books - and have a lot of experience - to translate the 'facts' to 'knowledge'.

For example: a lot of people read the docco and walk away with the impression that Buffer Cache Hit Ratio is a major tuning tool. The authors of the better tuning books almost invariably prove that, while it can be used as an indicator of mis-tuning, it can rarely be a reliable tuning tool in itself.

(In fact, since 8i, we have been teaching that BCHR should be the 'last thing to tune, after all other tuning has been accomplished'. And some of us - myself included - think it is a waste of time, mainly because it causes complacency and diverts people from the real task. Papers and code what been published allowing a DBA to display any BCHR they desire at any time, proving (to me at least) that it is not a reliable indicator.)

Since you ask specifically about performance tuning, you need to be aware that tuning is a dynamic process - ongoing and context sensitive. You simply can not tune a system once and forget it. Therefore, understanding what things invalidate previous tuning, and how to develop your own comfortable tuning method[ology] based on internals is beneficial. This can have a very steep learning curve, and the cited books (especially the cited authors) have an approach to presentation that can assist in that learning curve.

All of those authors are specialists in extracting the best performance possible out of Oracle. Oracle's documentation only provides guidelines for performance tuning, at best, and does not generally have case studies.

I personally have found that the cited books have information that makes a huge difference in being able to tune applications in great depth, often saving hundred of thousands of dollars in license fees and speeding up the tuning process.

Most important is that Oracle provides very few simple guidelines to do-it-yourself test cases. they assume you will either use the full toolset, or build your own tools. Books like those, as well as Tom Kyte's "Expert Oracle Database Architecture: 9i and 10g Programming Techniques and Solutions" provide simple tools, examples how to use them, and understanding how to interpret the results.

-- 
Hans Forbrich   (mailto: Fuzzy.GreyBeard_at_gmail.com)   
*** Feel free to correct me when I'm wrong!
*** Top posting [replies] guarantees I won't respond.

Received on Fri Jun 30 2006 - 09:38:38 CDT