Re: Slow retrieval from freetext column

Lex,

Our customers report that OMNIDEX is much faster and easier to maintain than interMedia Text indexes. In one benchmark, we brought a 13 minute query down to a few seconds. Also OMNIDEX works equally well on both structured and unstructured data, on both high and low cardinality columns, on both binary and character data, and doesn't require a CLOB data type.

We would be happy to estimate the performance improvements that OMNIDEX could provide for you (see http://www.disc.com/eval.html). If you would like more information, please contact me.

Cheryl Grandy
DISC
cgrandy_at_disc.com
303 444-4000

In article <8j01sq$6jq$1_at_nnrp1.deja.com>,   Lex <mlecza_at_amsnet.com> wrote:
> Why would I want to use this instead of interMedia Text indexes?
>
> In article <8it9l5$5lu$1_at_nnrp1.deja.com>,
> cgrandy_at_disc.com wrote:
> > One option to look into is the keyword indexing of OMNIDEX, which
 can
> > layer on top of Oracle and perform %string% searches instantly, no
> > matter what size the table. See http://www.disc.com/keyword.html
 and
> > http://www.disc.com/artindex.html or contact me for more
 information.
> >
> > Cheryl Grandy
> > DISC
> > 303 444-4000
> > cgrandy_at_disc.com
> >
> > In article <39505F88.71AB4C83_at_lib.monash.edu.au>,
> > kate.roberts_at_lib.monash.edu.au wrote:
> > > Dear all:
> > >
> > > My problem resides with the slowness in retrieving results when
> > > searching on one column of one table.
> > >
> > > The query ( select * from tablename where columnname like '%
 string%')
> > > searches a column containing large blocks of text, and tries to
 match
 a
> > > string in that text-block.
> > >
> > > I've been advised that indexing [using create index indexname...]
 on
> > > Oracle would be useless, since this would index each textblock as
 a
> > > left-justified single entity... rather than indexing each word/
 phrase
> > > in the text-block.
> > >
> > > I have 100,000 records in my table, (table-size is 200mb) and at
 40,000
> > > records the search took 12 seconds. At 100,000 records it takes
 70
> > > seconds.
> > >
> > > I have dropped and recreated the table, this time making the
 initial
> > > extent more than 200mb. Still no improvement.
> > >
> > > Does anyone know of a way to create an index on words/ phrases in
> > > textblocks- in a "text" column??????
> > >
> > > Or can anyone suggest another way to hasten the retrieval
 process????
> > >
> > > Regards,
> > >
> > > Kate Roberts
> > >
> >
> > --
> > Cheryl Grandy
> > DISC
> > Get OMNIDEX for the fastest
> > applications ever
> >
> > Sent via Deja.com http://www.deja.com/
> > Before you buy.
> >
>
> Sent via Deja.com http://www.deja.com/
> Before you buy.
>

--
Cheryl Grandy
DISC
Get OMNIDEX for the fastest
applications ever


Sent via Deja.com http://www.deja.com/
Before you buy.