Runtime filesystem consistency checking

I thought the other poster's point was about rotational speed. If the disk rotates at 100 revolutions per second then you may have to wait ten milliseconds in the worst case, even if the head is already positioned correctly. That ten milliseconds is not getting any shorter because disks are not spinning faster. However, the other components in the system are getting faster, so the ten millisecond overhead becomes more and more significant. Similarly, the disk head takes almost as long to move into position today as it did twenty years ago, even though processors and RAM are many times faster.

Or maybe the point is that larger filesystems necessarily require more random accesses and hence more disk seeks when you fsck them. Larger RAM would mitigate this but I don't know whether increased RAM for caching has kept pace with filesystem sizes enough. An fsck expert would be able to give some numbers.