[EMAIL PROTECTED] wrote:
"Igor Sereda" <[EMAIL PROTECTED]> wrote:
We have a database that can possibly grow into millions of rows. Some =
tables have TEXT fields, which may store texts of signigicant length. =
All other data is mostly numeric values.
We have a thought of moving all large texts into a separate table, and =
replacing text_column with text_id in the rest of the schema. The =
assumption is that db pages are allocated fully to a single table, so =
the numerical part of the database will end up in a few db pages and so =
we'll be able to quickly run queries over them. (We won't have queries =
for texts, only look-ups by text_id.)
Is our assumption correct? Is that a pattern someone here has =
implemented maybe? How does the size of the whole database affect =
queries to a single table?
This is a good assumption. Keeping large CLOBs and BLOBs
in a separate table and referencing them by rowid is what
I do.
I've had good luck with putting large texts into their own files, and
storing the file names in the tables. But that precludes searches
within the texts themselves. My applications didn't need to do that.
b.g.
--
Bill Gatliff
[EMAIL PROTECTED]
-----------------------------------------------------------------------------
To unsubscribe, send email to [EMAIL PROTECTED]
-----------------------------------------------------------------------------
