Re: Oracle Datafile size in Windows NTFS

"Nuno Souto" <nsouto_at_optushome.com.au.nospam> wrote in message news:3dec8a0d$0$31536$afc38c87_at_news.optusnet.com.au...
> Tue, 3 Dec 2002 10:08:20 -0000, Niall Litchfield said (and I quote):
>
> > autoextend) to a size I was comfortable with. I picked 4gb - result one
dead
> > financial system :(. Similar bugs with 2gb and 4gb datafiles have
existed on
> > various operating systems and various versions of Oracle.
>
> Rule #1 of defensive IT: never size anything at a power of 2 boundary.
> Particularly if that boundary is near a hardware limitation.
>
> These are the dreadful "off-by-1" type of bugs. John Bentley, the grand
> dad of defensive programming, identified this one in one of his books
> around 25 years ago.
>
> Still as true now as back then. And I'm told the new programming
> languages are supposed to help avoid coding errors...
>
> ;)
>
> > about to balance IO. It also avoids any future bugs with a round 2gb
file
> > size, the only thing that might leap out would be that since I use
locally
> > managed tablespaces a size of 2gb+64k for the bitmap would make a little
bit
> > more sense.
>
> While we're on the subject, something that popped in my mind recently: if
> we use LMT and AUTOEXTEND, what happens to the bitmap as the datafile
> grows? Does it grow as well?
>

Hi Nuno,

Sound advice by all.

In answer to your question, Oracle will allocate additional bitmap blocks if necessary (although these will not be "next" to the initial set).

Cheers

Richard Received on Tue Dec 03 2002 - 07:11:21 CST