Oracle FAQ | Your Portal to the Oracle Knowledge Grid |
Home -> Community -> Usenet -> c.d.o.server -> Re: Re-use of Leaf Nodes
Douglas Hawthorne wrote:
> "Howard J. Rogers" <hjr_at_dizwell.com> wrote in message
> news:40a217c8$0$23828$afc38c87_at_news.optusnet.com.au...
> SNIP.
>
>>Oh lord! It never stops, does it?! Just when you think you've got >>something nailed, some other degree of freedom pops loose! >>
I have to confess my head is spinning from all of this stuff now. I think I'll go back and ponder the delights of block dumps for a while. You know where you are with them. Have you noticed that the increase in index size is about 10% of its final size? You're getting 9 block inflations of the index when your index ends up being 90 blocks big. I got a 368 block inflation when my index ended up being 3622 blocks.
And you only see this result when PCTFREE is 10%.
I think there's a relationship there, don't you? I wonder what happens when you build the index (initially AND at the end) with PCTFREE of, say, 30%? (Yikes. I can't ask questions like that and not find out, so OK, I did the test and we get an end-of-test index inflation of ...[wait for it].... 30% (a 480 block index became a 690 block one).
I still can't see why the index balloons by PCTFREE with a rebuild when the loading of records into the index should have been respecting PCTFREE in the first place anyway (except that the figures suggest it wasn't) .... My head hurts.
Richard, are you there??
Regards
HJR
Received on Wed May 12 2004 - 14:11:43 CDT