Re: Question on Installing Oracle.

Hello Howard,

Thank you for clearing that up but I only received the information from Chapter 3: Disk I/O and Fragmentation in the Oracle Performance Tuning Tips & Techniques. Maybe I misread soemthing so I am going back to the book to make sure I read it correctly. Again I only got this info from a book.

Thanks again,

Kev.-

On Wed, 16 May 2001 21:20:29 +1000, "Howard J. Rogers" <howardjr_at_www.com> wrote:

>First off, the pctincrease setting for SYSTEM is DEFINITELY 50 in 8.0,
>8.1.5, 8.1.6 and 8.1.7
>
>And to say that "50 or 100 will REDUCE fragmentation" is a profound
>misunderstanding of the true state of affairs, and is (let's be blunt here)
>completely and utterly untrue and wrong, and I am somewhat gobsmacked that
>it is possible to be SO wrong.
>
>Fragmentation is caused by odd-sized extents within a tablespace.
>Fragmentation can NOT ever happen when all extent sizes are the same within
>a given tablespace. PCTINCREASE *guarantees* odd-sized extents within a
>tablespace. It was daft idea to invent, and an even dafter idea to make
>it's default 50, and potentially an even worse one to make SYSTEM use the
>default (but I would never change it for SYSTEM, as I said).
>
>What you have posted is the *complete* opposite of the truth.
>
>PCTINCREASE should be zero for ALL segments, without exception -except for
>anything stored in SYSTEM (and that last qualifier, I only admit to
>grudgingly).
>
>And let's be clear: to say that you "need to reduce the number of extents"
>in the SYSTEM tablespace is complete and utter nonsense. *IF* the only
>things that are stored in the SYSTEM tablespace are the data dictionary
>tables (as you assert in another post) then you do NOT need to reduce the
>number of extents, and you are banging your head against a brick wall in
>thinking that you do. Stop trying to manage the SYSTEM tablespace better
>than Oracle does the job: I guarantee that unless you are in the top 0.001
>percent of DBAs in the world, you'll do a worse job, not a better one.
>
>HJR
>--
>=============================!!=============================
>The views expressed are my own only, and definitely NOT those of Oracle
>Corporation
>=============================!!=============================
>
>
>"Kev.-" <java2e_at_yahoo.com> wrote in message
>news:3b01d34b.16300609_at_ns1.nothingbutnet.net...
>> No, I believe they are set to 0. I know that 50 or 100 will reduce
>> the fragmentation but it is like pulling teeth to get anyone to
>> listen.
>>
>>
>> Do you think that because it is 0 that is causing most of the problem?
>> I know the with 0 the extents will be uniform in size thus reducing
>> wasted space. But again I need to reduce the number of extents.
>>
>> Thanks.
>>
>>
>> On Wed, 16 May 2001 02:00:59 +0200, Bruno Jargot
>> <bjargot_at_club-internet.fr> wrote:
>>
>> >On Tue, 15 May 2001 21:37:53 GMT, Kev.- wrote:
>> >
>> >>Hello Howard,
>> >>
>> >>>Who cares if it's too small? Autoextend is set to on by default (for
>> >>>SYSTEM), and so when it needs more space, you get it. You haven't been
 daft
>> >>>enough to stop it autoextending, have you? SYSTEM must never run out
 of
>> >>>space.
>> >>No I am not daft enough to do that. BUt when I query to see the
>> >>extents it shows that the system and catalog are well over 300
>> >>extents. This just seems like alot of extents to me and it is getting
>> >>worse by the day.
>> >
>> >One question :
>> >The pctincrease of the tablespace SYSTEM is 50, isn't it ? And all the
>> >segments in the tablespace SYSTEM have a pctincrease of 50 ?
>>
>
>
Received on Wed May 16 2001 - 09:16:04 CDT