Currently researching about this topic from brendan gregg’s book. Not sure if it is worth the hassle. It seems too niched topic.
Few things I can learn would be simply what UNIX uses etc.
But is the comprehensive knowledge of that helpful in performance tuning or troubleshooting purposes?
If anyone is wondering about it, here is a start to a rabbit hole. 
These terms were important 50 years ago, and are now about as relevant as ferrite core memory. (This was known as Little Old Lady memory, because it was hand-woven).
The CHS cylinder/head/sector index shows the physical design. All the heads move together horizontally (so they describe a cylindrical shape on the rotating discs). One of the heads is energised, so only one of the surfaces on the platter is selected (out of possibly 20). A particular sector is selected (using timer markings on the underside of the platter). And your 512 bytes is read or written.
As an added joy, the sectors were not necessarily consecutive, because the electronics were too slow to deal with that. Logical sectors were usually "interleaved", so instead of waiting a full revolution to be ready for the next sector, it might just skip one and be ready for the following data block.
CHS addressing might have been used in 1970 at the programming level, but I believe the logic would have been shifted into the controller card by 1973. There is no scope for performance tuning here, and Unix/Linux device drivers work to a standardised interface several levels above this. There might be some offline diagnostic access route for factory use.
Even by the 1980s, disk controllers would queue access requests and deal with them in “elevator mode” to minimise head movement: that is, sweep in one direction across the disk satisfying requests in cylinder order, then make the reverse sweep.
I first met Mel when I went to work for Royal McBee Computer Corp.,a now-defunct subsidiary of the typewriter company. The firm manufactured the LGP-30...