Average distance between two points (first is current location, second is target location) when both are uniformly distributed in [ 0 .. +1 ] interval is not 0.5, it’s 1/3. If the full platter seek time is 8ms, average seek time should be 2.666ms.
Here's a good paper that explains why the 1/3 number isn't quite right on any drives manufactured in the last quarter century or so - https://www.msstconference.org/MSST-history/2024/Papers/msst...
I'd be happy to answer any other questions about disk drive mechanics and performance.
Is "full seek" a synonymous for worst case time to reach a position occurring less than 1% of working time?
From the article: max seek time is 15.2 ms + additionally 0 to 8.3 ms of rotational latency.
Reordering of sector accesses by NCQ should reduce worst case scenario occurrences.
When I was reviewing it for publication I ran a couple of tests and found more like 18 on the devices I tested, but I’m sure there are some that do 15. 25 is probably on the slow end. (although I’ve never tested a HAMR drive - their head assemblies are probably heavier and more delicate)
Old SCSI 10K drives could hand a huge queue and reach 500 random read IOPS, sounding like a buzzsaw while they did it. Modern capacity drives treat their internals much more gently, and don’t get as much queuing gain. Note also that for larger objects the chunk size is probably 1+ rotations to amortize the seek overhead.
You can do this test yourself with fio —readonly and root access to a hard drive block device, even if it’s mounted. (good luck reading any files while the test is running, but no damage done) Pick a variety of very high and low blocks, and the min delay will be when rotational delay is close to zero.
That’s a bogus number from an ancient slide deck for a class in 2001 or so, that’s misled generations of folks googling for the answer.
Note also that the outer tracks are longer (google ZCAV) and hold more data, so seeks across uniformly distributed block numbers do not generate uniformly distributed track numbers.
But you can simplify this problem down and ask: with no loss of generality, if your starting point is always 0 degrees, how many degrees clockwise is a random point on average, if the target is uniformly distributed?
Since 0-180 has the same arc length as 180-360 then the average distance is 180 degrees. So average half-platter seek is half of the full-platter seek.