Erik E. Fair wrote:
| In one sense, disk technology has not changed much since Kirk did the work
| (we still have platters that spin with heads that seek), however, some
| things have changed (e.g. Fast SCSI, Fast & Wide SCSI, Ultra SCSI, IDE,
| EIDE, plus disks that spin at 4500, 5400, 7200, and higher RPM), so I
| wonder when the last time all the default parameters of FFS were revisited
| in light of those things about disk technology that have changed.

More importantly, today we have "notched" drives, automatic bad sector
forwarding, on-drive caching, and the like.

Which means:
- Trying to keep things on same cylinder is likely a waste of time due to
  notched drives and other block address translations, although the general
  concept of locality to minimize disk seeks is still valid.
- Interleave is probably meaningless.  Don't modern drives start sucking
  in a whole track wherever the head may be, then spit what you want back
  at you from the cache?  Having a non-zero interleave also implies many
  I/Os to fetch contiguous blocks of a file; the kernel overhead of that
  may be the predominant cost factor.

IMHO, with today's disks, a LOT of the "optimizations" that were quite valid
in the days of the Vax 11/780 are useless today, and the FFS code can do
with a few simplifications as a result.