Has anyone done testing on a mechanical drive of this size regarding performance with these different configurations?
Until a few months ago I had multiple mechanical HDs connected to my M1 Mac Studio and the Finder lag was horrible…and I suspected the "access" time (or "seek" or whatever it's called) of each drive was the culprit.
Were you filesystems APFS or HFS+?
Also were you running Sonoma?
I'm wondering if that access delay is different for APFS vs. HFS+.
In general HFS+ is better optimized for HDD. This includes data structures that keep data commonly accessed together close together on the disk and allocation algorithms designed to minimize fragmentation (which leads to more non-sequential access/random seeks which are orders of magnitude slower on HDD). All else being equal I would avoid APFS on HDD.
On the other hand, there have been reports that Sonoma, Finder in particular, has gotten slower with very large folders. This may be most visible with HDD.
Would multiple HFS+ partitions be slower than one partition?
If accessing one file at a time or working entirely within one folder, performance would likely be the same.
However, if one partitions an HDD into multiple partitions and then simultaneously accesses files from each partition, the HDD would have to seek to each partition to access each file. As the partitions force a certain distance between locations on the disk, it is likely this would lead to more and larger seeks than accessing the same files all on one partition. That is it would likely be slower.
It would be a unique situation that many small HFS+ partitions would be faster than one large HFS+ all else being equal.
In theory there may be a slight benefit to partition an HDD around it's "speed zones" if one didn't need all the capacity online at the same time. In general the "outer edge" (typically the beginning) of an HDD is somewhat faster than the "inner edge" (typically the ending).
So in theory, one could partition the first 2 / 14 TB for active use, the middle 10 TB for archive, and not use the last 2 TB. I/O to that 2TB active partition would focus I/O on the faster parts of the disk and reduce average seek times. Not sure how significant this would be practice and therefore worth the complexity. I would test/benchmark before committing to this.
How about multiple volumes versus one on APFS?
My guess is that creating multiple volumes in one APFS container is likely to lead to more fragmentation which is likely to get real slow in the context of an HDD. I would avoid all else being equal.
