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

Home -> Community -> Usenet -> c.d.o.tools -> Re: Opinons/Philosophy on adding flags to table to reduce queries?

Re: Opinons/Philosophy on adding flags to table to reduce queries?

From: Rolf Czedzak <520040398192-0001_at_t-online.de>
Date: Tue, 10 Jul 2001 11:34:11 +0200
Message-ID: <9iei9s$sku$02$1@news.t-online.com>

Agreed. But 'better' does not mean 'good'. You have to think of  pathologically sized rows to justify extra amount of resources for index access -be it bitmapped or not- in the mentioned case of 80% pass.
I'd try to divide into 'true' and 'false' parts of the original table, providing
a view '<true> union <false>' to gain access to all elements if needed. Just my 2 cents.

Alexander V. Silantiev <silantiev_at_bashkortostan.ru> schrieb in im Newsbeitrag: 9i97vn$js5$1_at_poikc.bashnet.ru...
> If cardinality of tested columns is low, it would be better to combine
 that
> columns in bitmap index (especially
> when the data in this table is fixed)
>
> Rolf Czedzak <520040398192-0001_at_t-online.de> ñîîáùèë â íîâîñòÿõ
> ñëåäóþùåå:9hsolg$3ek$07$1_at_news.t-online.com...
> > Would bitmap index lead to anything but 80% pass. I do think, that's far
 to
> > little selectivity to use an index.
> >
> > Niall Litchfield <n-litchfield_at_audit-commission.gov.uk> schrieb in im
> > Newsbeitrag: 3b419c55$0$8510$ed9e5944_at_reading.news.pipex.net...
> > > If the data is fixed would a bitmap index suffice?
> > >
> > >
> > > --
> > > Niall Litchfield
> > > Oracle DBA
> > > Audit Commission UK
> > > "Jeff Wilson" <jwilson2000_at_home.com> wrote in message
> > > news:Xgc07.288435$p33.5772965_at_news1.sttls1.wa.home.com...
> > > > What is the conventional wisdom regarding adding a boolean column to
 a
 table
> > > > that is simply the result of a test on the values in about 6 other
 columns?
> > > >
> > > > I'm dealing with large tables and my query currently involves
 conditions
 on
> > > > the values of these 6 columns. I index all of these columns, but it
 seems
> > > > like it would be much faster to simply look at a single boolean
 value.
> > > > Probably about 80% of the rows in the table pass the boolean test.
> > > >
> > > > The data in this table is fixed so I don't have to worry about an
 update
> > > > affecting theboolean.
> > > >
> > > > Would only a purist avoid something like this? Are there
 considerations
> > > > here I'm missing?
> > > >
> > > > Thanks.
> > > >
> > > > --
> > > > Jeff Wilson
> > > > jwilson2000_at_home.com
> > > > Eugene, Oregon
> > > >
> > > > (541) 684-8590
> > > >
> > > >
> > >
> > >
> > >
> > >
> >
> >
>
>
Received on Tue Jul 10 2001 - 04:34:11 CDT

Original text of this message

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