I agree with most of their points, but their assertion that "writing files to disk is one of the slowest operations you can do" is laughably wrong. File IO is optimized to the extreme and cached on many layers from the IO library all the way down to the hardware.

I assume the writer is only familiar with synchronous disk IO which is ironic considering the mention of asynchronous programming.

File IO is one of the fastest operations if done right and some popular databases even use the OS cache instead of their own, MS SQL server is the only database I'm aware that doesn't.

I seriously question the authors knowledge of what it takes to "scale" if they think file IO is going to slow you down