Modern hard drive write speeds -- they slow down as they fill up?

Boy you can really see the “slowing down as the drive fills” phenomenon illustrated in this 110 (full disk) copy graph… it’s going from 0 bytes filled to 12tb filled!

One other factor to consider is filesystem.

In my *nix-centric experience fs can be a much bigger limiter than the physical disk. I think last I looked the inner vs outer tracks were more like 30% write speed but for example xfs on linux can be > 3x slower for the last 10%. Dunno re: Windows.

However, if the inner-tracks are much better, then what some do (esp before SSD) is make a partition using those tracks for temp and force “short stroking”. Before SSDs, this was actually done more for IOPS (using just the inner tracks also forces the heads to have less distance to seek back and can increase IOPS by an even bigger factor than write speed with some disks/busses/OSs).

1 Like