>I have an index on all tables with the primary key, which is clustered. That index is the default one and does not have any other field combined nor any included fields.
>
>I just created a new index on the primary key with the inclusion of two fields to satisfy the optimization of one particular SQL select command. To my surprise, the execution plan does not pick it up. The execution plan still takes the main primary key index.
>
>How can I avoid that situation?

Quite often you can force a certain index using a function with nonessential info added in front of order of resolving - which in itself of course is minimally less performant as it causes unnecessary cpu cycles - like "!" + pk + field1 + field2 for a where-clause and changing the wish-to-be-used index likewise. If the targeted index shows perfomance better by a large amount, the cpu cycles generated by the "!" should not matter that much...

HTH

thomas