Re: general recommendations for setting block size?

Very good point, Thomas. Same reason as large cluster sizes waste disk. If you have a database with lots and lots of very small tables and a few large ones, it is quite possible that some of the 8K block size will be wasted for the small tables. A special case, agreed. But also one that strikes quite often, unfortunately. Particularly if the database is subjected to creation and drop of small temporary tables as the application(s) executes.

Then again as I said, with technology going the way it is going I wouldn't be surprised if this didn't change soon. Currently the only thing I can see that might be really counter-productive with the 8K size is the speed of transfer in multiple access: 8K * lots of I/O transfer makes for expen$ieve overall I/O bandwidth. This will change soon.

--
Cheers
Nuno Souto
nsouto_at_nsw.bigpond.net.au.nospam
Is there a nospam domain?
http://www.users.bigpond.net.au/the_Den Thomas Kyte <tkyte_at_us.oracle.com> wrote in message news:tq7XN8g2SCHjgHv3M72bwCFYD90k_at_4ax.com...
> A copy of this was sent to "Nuno Souto" <nsouto_at_nsw.bigpond.net.au.nospam>
> (if that email address didn't require changing)
> >
>
> Sorry -- i'll have to disagree that 8k uses more space then 4k.
>
> There is much less block overhead in an 8k database (almost 50% less).
> I can fit more rows on 1 8k block then on a 4k block in general.
> I'll have less chained rows for long rows (and chained rows consume more
storage
> then non-chained rows).
>
> You'll want to watch your pctfree and pctused since they are percentages (20%
of
> 8k is lots more then 20% of 4k) but other then that -- whenever i take a <8k
> blocksize database and put it into an 8k blocksize database -- it takes less
> space, not more.
>
>
> So -- why the theory that 8k wastes space?
>
>
Received on Thu Sep 09 1999 - 08:47:54 CDT