thanks for the article! The Extended Event is definately one I will be using in the future.
What I would like to point out though is something that Uwe Ricken touched on above and that is the subject of unequally filled datafiles leading to hotspots on highly transactional OLTP systems and page contention.
Although a new datafile is added, data is not automatically distributed evenly across them all as you have demonstrated. Automatic Distribution takes a relatively long time before equality is achieved and (to the best of my knowledge) the only way to force this to happen is to rebuild all clustered indexes across the database assuming of course clustered indexes have been configured on all the tables. Heap tables will not be data distributed across the new datafiles using this technique. This is something I would have liked to have seen described on your article.
Bear in mind though that this will require downtime to be planned in as is normal with the rebuild of a clustered index and on very large tables is going to take time to complete.
Yet another reason to plan the design of a database with foresight and the future in mind.....
Just my 10 pence worth.. 🙂