concurrent statistics gathering

anyone using concurrent statistics gathering, as described in https://blogs.oracle.com/optimizer/post/concurrent-statistics-gathering ? Any tricks / hints and tips?  

We are on Oracle 19.20, with large OLTP databases (many db objects, partitioned tables) on Solaris and Linux. Statistics gathering is not really "concurrent", as after some time there is no really concurrent processing anymore. Support is not really helpful until now.  

Thanks and Regards

willyk

--
http://www.freelists.org/webpage/oracle-l