Poor Query Optimizing

Hi there,

I've got a SQL 2005 table(DataTable) with 5.5 million rows and a clustered index on an integer column. The column is a foreign key pointing to another table (LookupTable) that associates the integer with an nvarchar(50) column. Both integer and nvarchar(50) columns are in a clustered index.

I'm running equivalent select statements with slightly different where clauses, and am not seeing the performance similarities I'd expect.

select * from DataTable join LookupTable on DataTable.Integer = LookupTable.Integer

1. Setting the where clause to the nvarchar(50) value on the LookupTable takes 64 seconds.
2. Setting the where clause to Integer = (select integer from LookupTable where nvarchar(50) = 'mysearchstring') takes 61 seconds.
3. Setting the where clause to Integer = 526 takes 1 second.

Is this poor design in SQL, or is it a misconfiguration? I would have expected SQL to modify my query to essentially be #3 in all cases before touching the DataTable, which would cause all three versions to take 1 second.

Thanks,
Jason