Thanks for the article, quite interesting.
Still, it does not mention any word about the files/filegroups and eventually partitioning that is generally used behind the scene on big SQL Server "on premises".
Having (at minima) one file per CPU is best practice. Is there any "easy command" to take a backup defining a number of targeted files, or at restore time without any need to script all this "at posteriori" ?
What would be your recommendation for people wanting to test the cloud RDS and compare on premises fat DBs ?