I am in favor of separating the large fact tables into their own filegroups - based on better data and analysis.
I don't necessarily think that the fact tables should each have their own filegroup. But if you can split out the fact tables into some configuration of new filegroups - it would prove beneficial.
Where is the benefit of doing that?
With data this size, it provides a more robust opportunity for recovery. Sure you can see some performance gain (whether on same disks or not). But the big benefit is being able to take a filegroup backup and restore a filegroup backup (when/if necessary).
Maybe somebody wants/needs a 2TB group of tables restored to a dev or qa environment. With FG backups, you can achieve that without having to concoct a process to move the tables via bcp or ssis, or worse via backup/restore of the entire 50TB database.
Jason...AKA CirqueDeSQLeil
_______________________________________________
I have given a name to my pain...MCM SQL Server, MVP
SQL RNNR
Posting Performance Based Questions - Gail Shaw[/url]
Learn Extended Events