> Create a view in your relational data warehouse which will filter out
> inactive rows. Then create a dimension off the view instead of the table.
> You can have a view as source of rows for your dimensions and cubes.
>
> If you maintain indexes on your source tables, then every time you do the
> inserts, your indexes have to be updated as well, resulting in poor
> performance. Also if you have logging enabled for the database that serves
> as a dsata source for the cube, then every time you do deletes and
> inserts,
> the transaction log has to be updated as well.
>
> If your data warehouse is updated infrequently, you can set database
> loggin
> to "Simple" which would mean the only restore option available will be
> from
> a complete database backup -- this would minimize logging and make your
> inserts/deletes faster.
>
> Regards,
> Ilona Shulman
> Senior Development Consultant, DBA
> SSE Inc
>
http://www.sseinc.com >
>
> "Jéjé" <willgart@_A_hAotmail_A_.com> wrote in message
> news:OwKgtyjqEHA.3416@TK2MSFTNGP15.phx.gbl...
>> Hi,
>>
>> I want to delete "Inactive" rows in a dimension in my DW, but its slow
>> due
>> to relationships.
>>
>> First. I want to keep these relationships to insure a excellent data
>> integrity
>> Second. I delete rows linked to these deleted items before I delete my
>> dimension.
>> (So, normally, my "inactive" flagged rows will don't have any data in my
>> fact tables)
>>
>> Removing 100 rows in my dimension takes 3minutes !
>>
>> SQL Server appear to lock all the ralated tables and put a lot of
>> information in the log of the database.
>> So how can I improve this?
>> what are the best transaction options?
>> Maybe some table hints can be used?
>>
>> Currently, I'm looking for deleting the relationship just during this
> delete
>> step, then recreating these relations after the delete.
>>
>> thanks for your help.
>>
>> Jerome.
>>
>>
>
>
