FreeLists / oracle-l / Re: oracle block size = file system buffer size

[Joe Armstrong-Champ <joseph.armstrong-champ@xxxxxxxxx>]

Mark,

I actually got the info from Steve's site. I was just wondering what 
other people thought about this. I was thinking of increasing the block 
size on our indexes because I had also heard that this could reduce IO 
by flattening out the indexes. Sounds like keeping the block size equal 
to the file system buffer size outweighs the potentail advantages gained 
by the larger index block size.

Thanks.
Joe

Bobak, Mark wrote:

> Joe,
> 
> It's definitely a good idea for blocksize = filesystem block size.
> 
> Multiple blocksizes are there for transportable table support.
> 
> Steve Adams has lots of stuff on it here:
> http://www.ixora.com.au/tips/buffered_block_size.htm
> 
> Note that this is useful for databases NOT using directI/O.
> 
> Check out the link to Steve's article.  He does his usual
> excellent job explaining the concepts.
> 
> -Mark
> 
> -----Original Message-----
> From: oracle-l-bounce@xxxxxxxxxxxxx
> [mailto:oracle-l-bounce@xxxxxxxxxxxxx]On Behalf Of Joe Armstrong-Champ
> Sent: Thursday, September 30, 2004 4:32 PM
> To: oracle-l
> Subject: oracle block size = file system buffer size
> 
> 
> I have read that the oracle block size should be the same size as the 
> file system buffer size (when not doing direct io). If this is true then 
> what is the point in being able to have multiple block sizes? Is this 
> feature only helpful for databases using direct io?
> 
> Joe
> 
> --
> http://www.freelists.org/webpage/oracle-l

--
http://www.freelists.org/webpage/oracle-l