Oracle FAQ | Your Portal to the Oracle Knowledge Grid |
![]() |
![]() |
Home -> Community -> Usenet -> c.d.o.server -> Re: Count(*) is very slow
First thing, your RAM is toooo low for Oracle.
Oracle suggest to user around 1/4 of the RAM for the SGA. Is that mean that you are going to create a SGA for around 25MB?!? That is tooooo low!
Second, don't use the default parameter of init.ora. Try to tune the parameter like db_multiblock_read_count, db_buffers or things like that. But subject to the RAM that you have. I don't think you have a lot of room to tune it up.
Winnie
certified DBA
Andreas Kyritz wrote in message <70fiqs$pce$1_at_news.ipf.net>...
>Hi George,
>
>George Meltser schrieb in Nachricht <70ff74$qol$1_at_as4100c.javanet.com>...
>>I have a table with 3,5 mln rows and after it's analyzed it takes 3
seconds
>
>on which server-system do you get this result ?
>
>>to do
>>a count(*), so please stop blaming dbms.
>
>Sorry, but it really takes 36 seconds (3 seconds CPU-time on the server !!)
>after analyzing the table. There is, except of the operating system, no
>other proccess on the server. On the same machine the results with MS SQL
>Server, Centura, Informix and DB/2 - dbms where never so bad.
>
>The database I've used, was created with default values. Do you know, which
>parameter can tune this query ? Is it better, to work with a clustered
>primary key ?
>
>
>--
>Andreas Kyritz
>AKyritz_at_easy-soft-dresden.de
>
>
Received on Tue Oct 20 1998 - 00:54:07 CDT
![]() |
![]() |