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 performance config question

Re: Oracle performance config question

From: Kevin Brand <kevin.brandx_at_tel.gte.com>
Date: Wed, 4 Oct 2000 11:23:13 -0500
Message-ID: <8rflg4$t3v$1@news.gte.com>

I'm not sure how big your online redo logs are, but I can tell you from your init.ora parameters, that you will do a checkpoint after every 5MB worth of redo writes ( log_checkpoint_interval = 10000 ). I like to set this parameter so that the only time a checkpoint occurrs is at log-switch-time  when a redo log fills ).

Check your alert log for the number and frequency of checkpoints. My guess is that you are doing checkpoints way too often as Insert activity will write quite a lot of redo.

You may also want to increase the size of your online redo log members depending on their current size in relation to the parameter setting above.

Adjusting the frequency of commits in the "Insert" code so that you stay within a reasonable range of Rollback usage without causing too many extends/shrinks, may also improve performance ( that is only if you are issuing a commit after only 1 or just a few inserts ).

You may also consider using array inserts in your code, although this can become difficult if you have the occasional failed insert.

-Kevin
<mr_potato_head_at_my-deja.com> wrote in message news:8rdr6i$n0u$1_at_nnrp1.deja.com...
> Hi,
> I have an Oracle 8.1.6 database that is populated with over 3 million
> inserts a day. My problem is that my application is on 2 different
> servers. Server1 is used to collect data from all sources and then it
> transmits out parsed data via tibco's rendezvous. Server2, on the same
> subnet as server1, is listening for the broadcast traffic coming from
> server1 and then populates it's database with this data. I am able to
> throttle the broadcast data coming from server1, so server2 won't get
> overwhelmed, but during peak periods, I still get gaps because
> server2 can't populated it's database fast enough and the rendezvous
> queue overflows, so the data is lost. I'm running this on an E4500 w/
> 2gb of ram and a swap space of 4gb. I'd like to know what my
> configuration should be for this type of setup so I can get the best
> performance on my inserts.
> Here's what's in my /etc/system:
>
> set shmsys:shminfo_shmmax=1000000000
> set semsys:seminfo_semmni=256
> set semsys:seminfo_semmns=2048
> set semsys:seminfo_semume=64
> set semsys:seminfo_semopm=100
> set semsys:seminfo_semmsl=100
> set semsys:seminfo_semvmx=32767
> set shmsys:shminfo_shmmin=1
> set shmsys:shminfo_shmmni=100
> set shmsys:shminfo_shmseg=10
>
> And here's some variables in my init.ora:
>
> open_cursors = 100
> max_enabled_roles = 30
> db_block_buffers = 12162
> shared_pool_size = 66307968
> large_pool_size = 314572800
> java_pool_size = 0
> log_checkpoint_interval = 10000
> log_checkpoint_timeout = 1800
> processes = 315
> log_buffer = 163840
> db_block_size = 8192
> sort_area_size = 65536
> sort_area_retained_size = 65536
>
> thanks in advance...
>
>
> Sent via Deja.com http://www.deja.com/
> Before you buy.
Received on Wed Oct 04 2000 - 11:23:13 CDT

Original text of this message

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