New external SMR drive: second Read Test much slower than first

[shieldedtalisman]

Hi,

bought a new 2 TB WD My Passport external USB hard drive. I believe it's an SMR one.
Model: WDC WD20JDRW-11C7VS0

I did an initial read test when the drive was empty, and it gave a steady and stable read speed of approx. 260mb/s throughout the whole test.

Now I wrote some data on the drive, approx. 1 TB.

I ran another read test, and now the speed was slowly and steadily decreasing with time from approx. 110mb/s in the beginning to 60mb/s towards the end, then getting faster at the end, slower, faster.

The second read test took 3x the time of the first one, and there's only one day between them. But the 1TB of data written, also.
I also did a full reinitialize surface test after the first read test, so that was 2TB of writing data on it also.

Is this normal? I did accidentally move the hard drive while it was operational sometime after the first read test (but no active reading / writing occuring, and the move was sideways on the table, not a drop). The drive is so light, it moves a lot as soon as you touch the USB cable.
Everything seems to still be working fine, just want to make sure the drive is okay and operating how it's supposed to.

Sequence was: 1st read test, full reinitialize surface test, then wrote 1TB of my own data on it, later accidentally moved it while powered on and connected, then later second and much slower read test.

Thank you.

[hdsentinel]

Yes, what you see is completely normal and expected situation.

Generally when the drive was new, empty and you started the read test (even if sounds surprising) the drive did not read the data sectors: the drive has an internal look up table to "know" which sectors are empty (= all).

So before data actually written to the drive, it may show very fast read speeds - because the drive "knows" that it can respond the empty sectors so quickly, without actually reading the disk surface (like reading from a cache or so).

This trick is generally used for SSDs but yes, many SMR hard disk drives also work similarly. Such SMR hard disks usually support Data Set Management / TRIM commands (the Information page may confirm that) which is generally designed for SSDs to "notify" the drive when a file is deleted to free up the actual data sectors. If you see Data Set Management and TRIM functions supported on the Information page of the drive, then yes, it would indicate that the drive is SMR.

As soon as you write to the appropriate sectors (or fill the drive completely) then yes, the read test will (as then the drive really needs to access the data sectors) show slower results. This is possible with an SMR drive in general. It does not mean that the drive will be slower - just now you see the real transfer speed.

Constant 260 MByte/sec would be too good to be true for a 2.5" external drive, the 100-120 MByte/sec (and slower towards the end of the disk surface) is more realistic.

[shieldedtalisman]

Thank you very much for your quick reply and the great and very interesting information.
Yes, I did think 260mb/s sounds impossible and that it must be somehow related to SMR.
Yes, Data Set Mgmt. and TRIM are supported per the information page.

Do you have any advice how to deal with an 2.5" external drive that was moved/jerked sideways for ~10cm while it was on and connected, but not actively writing/reading? Didn't hear any bad sound during the movement and all the tests and behavior seem good and I did spot-check some files with SHA-256 comparison to original file on another drive. But no worries if this falls outside the scope of this forum and your help. It's just so very light, didn't realize you can't brush/touch the USB cable at all without moving it [in my setup].

Lastly, I really appreciate your program. It works great so far. I also read in another thread that you made an export configuration feature, which seems really handy in case of an OS reinstall.

Thanks again and have a good weekend.

[hdsentinel]

10 cm movement should cause no troubles especially if there was no read/write operation.
If the drive may dropped ("knock") to the table, then it could be a problem, but I suspect there should be no issue now.

Thanks so much for your kind words and support, really great honour

[shieldedtalisman]

Thank you again, appreciate it.