Execution plan surprise

Hi,

I have 2 tables: one with PK (+ Clustered index) on 3 columns and second with PK, FK on 3 columns (+ clustered index).

So I join those tables on 3 columns. I want to do some improvement and add 1 colum (of int type) to each table and create relation one-to-many. So now i may join on 1 column.... I have indexes on those new columns of course.

But I was surprissed but that join is slower ... On execution plan is showed that querry engine doesn't search newly added indexes, but old ones. And uses hash join (previously merge join).

Where is problem? Why sql server doesn't work on new indexes? Because they aren't clustered and not PK ?

Best regards,
Walter

[715 byte] By [WalterLuszczyk] at [2007-12-25]