I just had a situation where I had to include a column in an index. This required dropping the index, and recreating it. This got me thinking, surely that's a lot of unnecessary work.
Say I had instead created the new index first, then dropped the old one. Let's also assume that I have some way of indicating that the old index will not be dropped until after the new one has been created.
Would the server get any performance benefit by using the old index while creating the new one?