Move reportupdater reports that pull data from eventlogging mysql to pull data from hadoop via hive queries or even be transformed into hive-to-druid jobs that surface data on superset. We need to look at every report to see what is the best solution. We also need to have in mind that druid might hit a limit when it comes to the number of datasources.
How about joins between analytics replica data and eventlogging data? (those can only happen monthly but seems we do not have any of those right now)