Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Usenet -> c.d.o.server -> Re: oracle 7.3 server: performance problem with new dual proz hardware

Re: oracle 7.3 server: performance problem with new dual proz hardware

From: Alan Campbell <ajc_at_nospam.cgocable.net>
Date: Fri, 13 Nov 1998 22:27:45 -0500
Message-ID: <364CF8AF.11CAE936@nospam.cgocable.net>


You server may be thrashing doing paging operations. May I suggest the following.
You have 512MB of real memory, leave 128MB of that for the O/S, now you have 384MB to use for Oracle(if there are no other applications running on the server)..
I dont know your block size but use maybe 200MB for db block buffers which would be a value of 50000 if you have a block size of 4096. and try only 35000000 (35MB) instead of 350000000 (350MB)for your shared pool. you haven't indicated any info about file placement. If you used the default install, all datafiles are on the O/S drive. spread the files over the drives, and use performance monitor to see what is happening with i/o and your processors.

Ein Prosit
Al
ajc [at] cgocable [dot] net
uwe fechner wrote:

> hello,
>
> we are running a small, 5g big, database with oracle 7.3. under nt server
> 4.0 sp3.
> we first used an p133 with 64MB. this was slow.
> now we use an dualproz pentium II with 300MHZ, 512 MB and 2 9gig scsi
> drives.
> the database server is connected to the web/app server by an seperate
> fastethernet line.
> this is slower as the old configuration.
> nt has with the new hardware nearly 5 - 10 times the performance of the old
> hardware.
> oracle detects the 2 prozessors.
> we increased the amount of shared memory
>
> here is a part of the config:
> compatible = 7.3.0.0.0
> db_file_multiblock_read_count = 32 #
> LARGE / MOG
> db_block_buffers = 4500 # MOG
> shared_pool_size = 350000000 #
> feu/mog 14.10.98
> log_checkpoint_interval = 10000
> dml_locks = 500 #
> LARGE
> log_buffer = 8192 #
> INITIAL
> sequence_cache_entries = 100 #
> LARGE
> sequence_cache_hash_buckets = 89 #
> LARGE
> max_dump_file_size = 10240 # limit trace file size to 5 Meg each
> log_archive_start = true # if you want automatic archiving
> log_archive_dest = c:\orant\archive
> # define directories to store trace and alert files
> background_dump_dest=%RDBMS73%\trace
> user_dump_dest=%RDBMS73%\trace
> db_block_size = 2048
> snapshot_refresh_processes = 1
> remote_login_passwordfile = shared
> text_enable = true
>
> any one an idea ?
>
> greetings from germany
> uwe
> --
>
> Uwe Fechner
>
> ¤º°`°º¤ø,¸¸,ø¤º°`°º¤ø,¸¸,ø¤º°`°º¤ø,¸¸,ø¤º°`°º¤ø¤º°`°º¤ø,¸¸,ø¤º°`°º¤ø,¸¸,ø¤º°
> `°º¤ø,¸¸,ø¤º°`°º¤ø
>
> u.fechner@integrata.de http://www.integrata.de
> +49 211 96 555 47 Fax: +49 211 96 555 99
> uwe.fechner@pironet.de http://www.koeln.netsurf.de/~uwe.fechner
>
> ¸¸,ø¤º°`°º¤ø,¸¸,ø¤º°`°º¤ø,¸¸,ø¤º°`°º¤ø¤º°`°º¤ø,¸¸,ø¤º°`°º¤ø,¸¸,ø¤º°`°º¤ø,¸¸,
> ø¤º°`°º¤ø¤º°`°º¤ø,
Received on Fri Nov 13 1998 - 21:27:45 CST

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US