How FriendFeed uses MySQL to store schema-less data

A post from one of the FF engineers about how they have a schema-less data store and then build separate indexes to allow for fast querying.