Hi,

I'm looking for a good solution to remove unused records in my dimensions.
My users dislike to see unused objects in cube's dimensions and other
prompts in reports.

But the delete command is very slow due to a ton of constraints (to insure
good data quality).

so, any idea? guide?
maybe I can change my database to 1 user only during the delete step?
does I have to change and setup some isolation hint in the delete command?

thanks.

Jerome.

How are you going to identify unused rows?
Are your dimensions views over base tables? If == Yes then change the view
definition

DELETE is slow because it logs every row of the delete. Indexes etc also
need to be juggled around

--

Allan Mitchell MCSE,MCDBA, (Microsoft SQL Server MVP)
www.SQLDTS.com - The site for all your DTS needs.
www.Konesans.com


"Jéjé" <willgart (AT) BBBhotmailAAA (DOT) com> wrote

I'll identify these rows by executing a lot of queries against "all" my
tables to validate if my row is used.

For example.
my "calendar" table generates dates from 1990 to 2006
but , for the moment I've only data from 2002 to 2005
So, I want to delete dates between 1990 and 2002.
This table is linked to 15 fact tables.

For the moment I don't use views to access my data to process my dimensions.
And doing a join with 15 fact tables starts to become a huge work!

But its one of the solutions to create views.
I also evaluate the option to add a column "used" where (during my DTS
loading process) I'll update this field to Y or N and then use the view
against all "Y" columns.

any other options?


"Allan Mitchell" <allan (AT) no-spam (DOT) sqldts.com> wrote