[smartmontools-support] Inconsistent long test times

[Vedant Lath]

Sorry, both seem to be SMR drives. My bad.

On Sat, Mar 13, 2021 at 9:57 PM Vedant Lath <vedant at lath.in> wrote:

> ST2000LM015 is an SMR drive while WD20SPZX is a CMR drive - that explains
> the large IOPS differences.
>
> On Sat, Mar 13, 2021 at 2:44 AM Kent Watsen <kent+smartmontools at watsen.net>
> wrote:
>
>>
>> > Is this level of variability normal?
>>> > Should I return all the WD drives?
>>>
>>
>> You can register your purchased WD drives at Western Digital  (
>> https://support-en.wd.com/app/account/dashboard).
>> When done, you can open a support case whether or not this is to be
>> expected. And ask for the reasons.
>>
>>
>> Good idea - done!  (Hopefully not a new spam vector)
>>
>>
>> And then share with us please ;-)
>>
>>
>> Will do!
>>
>>
>>
>> My first response from WD was:
>>
>> I really apologize for the delayed response. Upon discussing this case
>> with the senior team, they wanted you to run diagnostics using data
>> lifeguard software. If you still unhappy with the drives then we can get it
>> replaced.
>>
>>
>> However, after pointing out that I don’t have a Windows or Mac computer
>> to test against and also that the question isn’t whether the drives are
>> functioning properly, but *why* the SMART test times vary so wildly
>> for drives having the same model-number, and if these test-time numbers are
>> normal, the response I received was:
>>
>> Thank you for your continued response. We can only comment on the behalf
>> of our tools and our tools (Windows data life guard diagnostics) only run
>> on windows and we do not have any information about Solaris.
>>
>>
>> That said, using `fio` to test the performance of the various drives, I
>> found that both the WD20SPZX drives outperformed the ST2000LM015.  The
>> results below only show the “write” test results because the ZFS L2ARC
>> cache validates the “read” values.  Note that, in all cases, the ZFS pool
>> contains just the single drive.
>>
>>
>> • For a Seagate Barracuda (ST2000LM015) drive, which had a 330 minute
>> long-test wait time:
>> • 4k
>> • Random write: IOPS=24, BW=97.6KiB/s (99.9kB/s)
>> • Sequential write: IOPS=21, BW=85.2KiB/s (87.2kB/s)
>> • 64k:
>> • Random write: IOPS=12, BW=785KiB/s (804kB/s)
>> • Sequential write: IOPS=12, BW=788KiB/s (807kB/s)
>> • 1m:
>> • Random write: IOPS=9, BW=9439KiB/s (9665kB/s)
>> • Sequential write: IOPS=9, BW=9930KiB/s (10.2MB/s)
>>
>> • For a WD Blue (WDC WD20SPZX-75U Revision: 6712) drive, which had a 737
>> minute long-test wait time:
>> • 4k
>> • Random write: IOPS=33, BW=136KiB/s (139kB/s)
>> • Sequential write: IOPS=34, BW=137KiB/s (140kB/s)
>> • 64k:
>> • Random write: IOPS=34, BW=2207KiB/s (2260kB/s)
>> • Sequential write: IOPS=34, BW=2215KiB/s (2268kB/s)
>> • 1m:
>> • Random write: IOPS=34, BW=34.5MiB/s (36.1MB/s)
>> • Sequential write: IOPS=35, BW=35.6MiB/s (37.3MB/s)
>>
>> • For a WD Blue (WD20SPZX-22U Revision: 1A01) drive, which had a 119
>> minute long-test wait time:
>>
>> • 4k
>> • Random write: IOPS=34, BW=138KiB/s (141kB/s)
>> • Sequential write: IOPS=35, BW=141KiB/s (145kB/s)
>> • 64k:
>> • Random write: IOPS=35, BW=2248KiB/s (2302kB/s)
>> • Sequential write: IOPS=34, BW=2191KiB/s (2243kB/s)
>> • 1m:
>> • Random write: IOPS=34, BW=34.4MiB/s (36.1MB/s)
>> • Sequential write: IOPS=35, BW=35.4MiB/s (37.1MB/s)
>>
>>
>> These results muddy any coarse of action.  Perhaps the WD20SPZX-22U
>> drives (from NewEgg) are best, having both better performance and smaller
>> “long-test” wait times.  If I could do it over again, I might buy all 10
>> drives from NewEgg in 5 separate orders (containing two drives each) spaced
>> a few days apart.  But, as it stands, I’ll just hold onto what I got and
>> live with the crazy-long test times on the WD20SPZX-75U drives.
>>
>> Thanks all for your comments.
>>
>> Kent
>>
>>
>>
>> _______________________________________________
>> Smartmontools-support mailing list
>> Smartmontools-support at listi.jpberlin.de
>> https://listi.jpberlin.de/mailman/listinfo/smartmontools-support
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://listi.jpberlin.de/pipermail/smartmontools-support/attachments/20210313/c61930d4/attachment-0001.htm>