I'm not questioning the validity of the data point, but why would RDBMS have a cost of 10 times per unit of storage? I haven't heard this reasoning before. Anyone know where this comes from?
If you use things like mongodb and co, you are giving up on single-node durability, for example, and you don't need to have as much efficient IO (e.g. reasonable fsync performances). Otherwise, I don't see why mysql would require more expensive hardware mongodb