• I really don't get it...

    10-20-30 ip uses the lazy spool so it takes several seconds.

    40-50-60 ip uses the clustered index seek so it executes in a flash.

    70 and more also uses the lazy spool so it takes more than a minute.

    Is there a way to force the use of the index instead of the costly table spool?

    By the way, it's a clustered index SCAN with the table spool instead of a clustered index seek. What's the difference?

    Thanks

    Stephane