Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Usenet -> c.d.o.server -> Re: Reverse key Indexes Oracle 8.0

Re: Reverse key Indexes Oracle 8.0

From: <markp7832_at_my-deja.com>
Date: Mon, 29 Nov 1999 20:07:24 GMT
Message-ID: <81umdp$1hg$1@nnrp1.deja.com>


In article <3842ac32.11741392_at_read.news.globalnet.co.uk>,   boulke_at_globalnet.co.uk (Keith Boulton) wrote:
> On 29 Nov 1999 12:46:39 +0000, Robin Smith
> <smithrc_at_zdbaora.nat.bt.com> wrote:
>
> >records will always be added to the right hand side of the index so
it
> >will become skewed. If you reverse the key then entries will also go
> >into the middle of the index so will become self balancing.
>
> A b*tree index cannot become skewed.
>

I believe that your comment that a b*tree index cannot become skewed is only true in theory. Depending on how the b*tree is implemented in code, activity unbalances the tree. In theory it should be rebalanced as activity takes place but there is a cost to doing so. This cost is highest for deletes and if my memory is correct the version 7 documentation claimed that Oracle rebalanced for inserts but it deferred much of the work for deletes. I have read some posts in the past questioning how good a job Oracle did rebalancing for inserts in the case of sequential keys. I can not remember the ver 8 documentation on this topic but I have not noticed any growth behavior changes on our indexes so I suspect there has been no major changes to the implementation of b*trees with Oracle. --
Mark D. Powell -- The only advice that counts is the advice that  you follow so follow your own advice --

Sent via Deja.com http://www.deja.com/
Before you buy. Received on Mon Nov 29 1999 - 14:07:24 CST

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US