Re: Why is NEXT_EXTENT changing all the time?

Walter T Rejuney wrote:

> Syltrem wrote:
> >
> > That is what Oracle DBA's say, and it also disagrees with what Oracle says.
> > Have a look at the last document on this page:
> > http://otn.oracle.com/deploy/performance/listing.htm
> >
> Wow!!!! It is like a lightbulb going on. Some of the stuff in that paper
> is not applicable to the type of work that I do, but I've saved that
> document and the next time I design a physical database I'll probably go
> by those concepts. I'm not sure I agree with the 160K/5120K/160M
> standards, but I'll probably settle on a single figure. On a basic
> level, though, what the author of the paper is saying makes consummate
> sense.
>
> Still, I have to wonder why Oracle allows PCT_INCREASE for tablespaces
> to default to 50% if not explicitly overridden. It is a nuisance and
> accounts for at least half of the emergency fix work that I have to do
> on various customer databases.

The sizes quoted above are absolutely unreasonable. The sizes should be based upon the block size of the operating system running Oracle as well as the number of blocks that O/S captures in a single read. Then everything should be multiples of that. The above numbers, for example, make no sense at all on my Solaris 2.7 machines: A 160K extent would be a waste. I do, however, follow a similar formula but it is based upon the individual application rather than some carved-in-stone numbers. For example a data-mart I just built has tablespaces sized at 256K, 25M, and 1G.

With respect to Oracle setting up that 50% default the only thing I can think of is that all large corporations occasionally do inexplicable things. This holdover from a long time ago should have dropped long ago.

Daniel A. Morgan Received on Fri Jan 26 2001 - 00:03:23 CST