I've had lots of trouble with this half-baked 'feature'.

A couple of dozen monitored instances ended up in an mdw db over 700GB data and 300-odd GB log.

Looking at the code inside explains some of the issues, it was probably written by trainees.

Make sure you assign good storage to both data and logs. This one caught me out, the SAN was not nippy and the clearing up wasn't happening properly on the datafile. Had to drop the whole thing twice and start over again with new defaults (choosing to retain less rather than more).

Watch carefully the size growth in the first few weeks and pre-allocate enough space. It can grow very large very fast.

Be careful, some snapshots are every 15seconds, that can lead to huge tables that are very slow to query afterwards. You may want to change them to 300 secs perhaps.

Also stagger the Agent jobs manually, you don't want all over them hitting the central db at the same time (clearly the defaults are not meant for multi-instance usage:hehe:)

There are bugs in the procs as well, worth doing an internet search.

Yes it comes free with SQL, but it is worth as much.:-D