Oracle FAQ | Your Portal to the Oracle Knowledge Grid |
Home -> Community -> Usenet -> c.d.o.server -> Re: Index management
Note in-line.
-- Regards Jonathan Lewis http://www.jlcomp.demon.co.uk The Co-operative Oracle Users' FAQ http://www.jlcomp.demon.co.uk/faq/ind_faq.html June 2004 UK - Optimising Oracle Seminar July 2004 USA West Coast, Optimising Oracle Seminar August 2004 Charlotte NC, Optimising Oracle Seminar September 2004 USA East Coast, Optimising Oracle Seminar September2004 UK - Optimising Oracle Seminar "Howard J. Rogers" <hjr_at_dizwell.com> wrote in message news:409334fa$0$25007$afc38c87_at_news.optusnet.com.au...Received on Sat May 01 2004 - 03:30:14 CDT
>
> After that I had indigestion with your 50-50 leaf block claim. Jonathan
> claims it's 50-50 *on average*, but not by volume, I think. Interesting.
>
I'm not sure I've ever been precise in my wording about how the 50/50 works - by default the tests I've done to date have always had uniform sized keys, so 50/50 by key count would tend to be 50/50 by volume, so I've never gone beyond saying just "roughly 50/50". This, of course, demonstrates, that for every test result and repeatable test case, there is always an enhancement to the test and refinement of understanding. (Richard's "50% by volume" is such an obvious detail to check - after it's been pointed out !! - it's also an extremely practical point to be aware of, some systems will have keys with lots of columns with lots of nulls, leading to all sorts of odd variation in key length). In fact, Steve Adams described a further detail to the split process when he was doing the Miracle Masterclass in Denmark in 2003, the preliminary target is a 'fair' split - which Richard now tells us is 50/50 by volume - but Oracle will then allow a little side-stepping on the split to mimimise the size of the necessary entries in the branch blocks.