Using SQL 2008 r2 and SSIS, I reload a fact table everyday. There are no PKeys, FKeys, or indexes on any of the tables. It is also not a relational database setup.
Records are extracted from TableA, transformed and either added to TableB or replace records in TableB. Once all the records are transformed I join TableB with TableC to load the fact table. I use an Execute SQL Task to issue the following statements.
>Truncate table fctTable;
Insert into fctTable
Select col1, col2, col3
From TableB, TableC
Where TableB.ID = TableC.id
There are no aggregates in the select, but there is some formatting of data. The result is 36 Million rows in the fctTable. This is correct. There is not one specific column that can be a primary key in this table. All column values can be in the table multiple times.
It is taking over an hour to load this fact table. Even though I have no keys or relationships of any kind, would an index help the performance? I have tried creating a view and selecting from the view but it doesn't improve performance.
There is no red speech bubble to format the code on my page.
Thanks!
paws27284
↧