Xref: alice comp.databases.oracle.server:30951
Path: alice!news-feed.fnsi.net!newspump.monmouth.com!newspeer.monmouth.com!rill.news.pipex.net!pipex!bore.news.pipex.net!pipex!not-for-mail
From: scooby <scooby.do@shaggynet.com>
Newsgroups: comp.databases.oracle.server
Subject: Re: Block Size and Checkpoints (NT)
Date: Sat, 10 Oct 1998 11:28:06 +0100
Organization: UUNET WorldCom server (post doesn't reflect views of UUNET WorldCom)
Lines: 44
Message-ID: <361F36B6.4607@shaggynet.com>
References: <6uv729$hn2$1@nnrp1.dejanews.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 3.01 (Win95; I)

fatman@pobox.com wrote:
> 
> Could someone help me with block sizes for Oracle 7.3 under NT 4? I'm
> quite new to both NT and Oracle, so please forgive me if the answers
> seem obvious.
> 
> I'm using a 4 GB disk (NTFS). I believe the OS block size (called
> 'cluster' I think) for disks of that size under NT is 4 kB, but I'm not
> sure. How do I find out?
> 
> The recommended value for Oracle's DB_BLOCK_SIZE parameter for small
> databases is 2 kB. Now I wonder, is it really ok to use a smaller DB
> block size than the actual OS block size (in my case presumably 4 kB)?
> 
> Lastly, I want my checkpoints to occur only at logswitches (that's
> actually why I started looking into block sizes). It seems there are
> two different parameters which achieve this:
> 
> a) setting LOG_CHECKPOINT_INTERVAL to a number larger than there are OS
>    blocks in the redo log file
> b) setting LOG_CHECKPOINT_TIMEOUT to zero.
> 
> Do I have to use both parameters or is it enough to set one of them? What's
> the difference between the two parameters/methods?
> 
> Thank you very much in advance
> 
> Fatman
> 
> -----------== Posted via Deja News, The Discussion Network ==----------
> http://www.dejanews.com/       Search, Read, Discuss, or Start Your Own

Regarding block size I run our NT based Oracle databases with a block
size of 2K with no problems. I have been told that the db block size and
os block size are not necessarily directly related under NT

Regarding checkpoints I made our LOG_CHECKPOINT_INTERVAL > Redo log size
and can confirm that the database only checkpoints at redo log switches.
I have never set the value LOG_CHECKPOINT_TIMEOUT so this must be at
it's default setting (whatever that is).

Hope this helps.

Regarding