When I bought a bunch of NVME drives, I was disappointed with how slow the maximum speed I could achieve with them was given my knowledge and available time at the time. Thanks for making this post to give me more points of insight into the problem.

I'm on the same page with your thesis that "hardware is fast and clusters are usually overkill," and disk I/O was a piece that I hadn't really figured out yet despite making great strides in the software engineering side of things. I'm trying to make a startup this year and disk I/O will actually be a huge factor in how far I can scale without bursting costs for my application. Good stuff!