|
then yeah it could be not useful to use index, because using index is possible to avoid SORT, however if the number of rows selected is almost all rows in the table and the index does NOT include all required columns, it then requires FETCH which may increase the cost a lot by using additional IO to get data...
so in this case, try to reorg data based on the join column, hopefully it will reduce the time used to sort data, and maybe creating clustered index on the join column is also good to have |
|