I just read about FriendFeed using MySQL as a data storage engine. Very interesting post that highlights how much people have come to trust MySQL. FriendFeed basically built their own object storage system inside a single table. They manually create and update indexes in other tables in a nicely distributed and sharded database server setup. Overall its what I’d call a “very impressive” implementation. I can’t help but think back to how BlogLines used (or perhaps still uses) BerkeleyDB to store their data without a traditional RDBMS storage engine. I wonder if FriendFeed considered using a solution like that?