There is no coping of data with the switch statement, the “switch” only changes the pointers. But if you have a merge or split involved you are definite to have a lot of IO and this approach needs to reconsidered under the new circumstances (You need to allocate time for downtime).

This approach will work best for databases on a single file group