Re: Tuning DB_BLOCK_BUFFER

On Thu, 17 Sep 1998 6:32:58 +0200, anonymous <anonymous_at_anonymous.com> wrote:

>
>On Wed, 16 Sep 1998 4:40:37 +0200, Winnie Liu wrote
>(in message <6tn86a$6nl_at_sjx-ixn4.ix.netcom.com>):
>>
>> HI everyone,
>>
>> Is it going to hurt the database performance if I increase the number of
>> DB_BLOCK_BUFFER to a large number? I am absolutely sure that the RAM and
>> SHMMAX parameter is set in a way that it is big enough for the whole SGA to
>snip
>> I know that you can set the db_lru_extended_statistic (or something
>> similar) to check the the performance increase for each addition of
>> db_block_buffer. But since that database is in production for 24*7. There is
>> no way I can shutdown the database and check for it. Does anyone have any
>> idea why this is the case?
>
>Be glad you did not use lru_extended_statistics on 7.3.4. ! Last week this
>caused me a total HANG while trying to get some usefull figures :( Oracle
>removed this feature from oracle-8 because of the troubles it caused.
>
>--
>Ronald
>
>---------------------------------------------------------------------------
>mailto:ronr_at_wxs.nl
>http://home.wxs.nl/~ronr/RonR.html (last update: june 8th, 1998)
>http://www.maccentral.com/news/9804/migration.shtml (forward migration home)
>

As long a you don't incur paging/swapping you should be OK. Make sure that you don't neglect the library cache sizing though, that's more important. Ensure it's big enough to avoid too many misses.

Regards,

Clive Bostock Received on Sun Oct 04 1998 - 04:12:06 CDT