Frank Steinmetzger wrote:
> Am Wed, May 07, 2025 at 09:18:16AM +0100 schrieb Michael:
>> On Wednesday, 7 May 2025 00:30:34 British Summer Time Dale wrote:
>>> […]
>>> I ran a hdparm test.  I wanted to see as accurately as I could what the
>>> speed was.  I got this.
>>>
>>>
>>>
>>> root@nas ~ # hdparm -tT /dev/sdb
>>>
>>> /dev/sdb:
>>>  Timing cached reads:   7106 MB in  2.00 seconds = 3554.48 MB/sec
>> These are rather pedestrian ^^^^ but I do not have any drives as large as 
>> yours to compare.  A 4G drive here shows this:
>>
>> ~ # hdparm -tT /dev/sda
>>
>> /dev/sda:
>>  Timing cached reads:   52818 MB in  1.99 seconds = 26531.72 MB/sec
>>  Timing buffered disk reads: 752 MB in  3.00 seconds = 250.45 MB/sec
>>
>> That's an order of magnitude higher cached reads.
> So you have a faster machine, possibly DDR5. Dale’s NAS is an old build. 
> That’s why it’s called cached.
>
> From the manpage of hdparm: This measurement [of -T] is essentially an 
> indication of the throughput of the processor, cache, and memory of the 
> system under test. This displays the speed of reading directly from the 
> Linux buffer cache without disk access.
> -------------------^^^^^^^^^^^^^^^^^^^
>
> The output of -t is the actual physical bandwidth. And for a big current 
> haddrive, 250 MB/s is a decent normal value.
>
>
> For comparison, this is from a nice SATA SSD (Crucial BX100 512 GB) on a 
> passive MiniPC with Celeron N5100 and DDR4 RAM (Zotac ZBox Nano CI331):
>
> root@schatulle ~ # hdparm -Tt /dev/sda
>
> /dev/sda:
>  Timing cached reads:   10950 MB in  2.00 seconds = 5480.12 MB/sec
>  Timing buffered disk reads: 1602 MB in  3.00 seconds = 533.56 MB/sec
>
>
> As you can see, the SSD is almost at the practical limit of SATA 3, which is 
> 600 MB/s. Wikipedia: Third-generation SATA interfaces run with a native 
> transfer rate of 6.0 Gbit/s; taking 8b/10b encoding into account, the 
> maximum uncoded transfer rate is 4.8 Gbit/s (600 MB/s). 
>
>>>  Timing buffered disk reads: 802 MB in  3.00 seconds = 267.03 MB/sec
>>> root@nas ~ #
>>>
>>>
>>> From what I've seen of other drives, that appears to be SATA 3 or the
>>> faster speed.  So, it is slow to respond but connects and works fine. 
> SATA2 runs at half of SATA3, at 300 MB/s. So even if your drive ran at 
> SATA2, you wouldn’t notice any impact in performance.
>


I have since did some more testing on this.  I ran the hdparm -t on
other drives on this same rig.  AM4 with AMD Ryzen 7 5800X and 128GBs of
G.SKILL DDR4.  Some of the hard drives are connected to PCIe SATA
adapter cards.  They pretty fast.  So far, the one with the fastest data
speed is the 20TB drive.  They look something like this. 


root@Gentoo-1 / # hdparm -t /dev/sdb

/dev/sdb:
 Timing buffered disk reads: 816 MB in  3.00 seconds = 271.83 MB/sec
root@Gentoo-1 / # hdparm -t /dev/sdc

/dev/sdc:
 Timing buffered disk reads: 576 MB in  3.01 seconds = 191.62 MB/sec
root@Gentoo-1 / # hdparm -t /dev/sdd

/dev/sdd:
 Timing buffered disk reads: 808 MB in  3.01 seconds = 268.87 MB/sec
root@Gentoo-1 / # hdparm -t /dev/sde

/dev/sde:
 Timing buffered disk reads: 730 MB in  3.01 seconds = 242.69 MB/sec
root@Gentoo-1 / # hdparm -t /dev/sdf

/dev/sdf:
 Timing buffered disk reads: 658 MB in  3.00 seconds = 219.30 MB/sec
root@Gentoo-1 / #


So, comparison of speed shows the 20TB is a little faster but sdd is
pretty darn close, a 16TB drive.  I currently have the drive in a
external enclosure that has a fan.  I have another drive in the exact
same model of enclosure.  The one with the 20TB drive runs at around 90F
and the fan spins faster.  The other enclosure with a older 6TB drive
runs at 87F with a slightly lower fan RPM.  The fans are temp
controlled.  So the 20TB runs a little warmer, about 3 degrees F
warmer.  It could be that it just has extra stuff packed in there and
that is normal or could it be that a little bit of the helium has
already got away.  Maybe??  Keep in mind, SMART showed a power up time
of only 2 hours when I first connected the drive.  Unless they can reset
the power on hours somehow, it should be a new drive.  It may have sat
on a shelf somewhere for no telling how long tho. 

In this newer enclosure, it seems to end up running at full speed,
despite the slow to respond and other info in the logs.  It concerns me
that it is slow to respond but it seems to work just fine once it gets
spun up and connects properly. 

I also found a option for smartctl that is interesting.  I checked the
output of SMART with the -x option.  It seems to include a lot of
different info than -a.  It is supposed to work better on m.2 sticks I
think but I haven't tested it yet.  I might add the -I option for hdparm
is also nice.  I think Michael mentioned it.  Some who are reading this
may want to check those options for those commands.  It may display info
that comes in handy. 

Over the past few weeks, I've ordered two 16TB drives and this 20TB
drive.  I may put this 20TB in my backup drive set.  It currently has
almost 40TBs of storage.  I have to split my video collection into two
pieces right now.  With this, I could have one large data backup drive
set instead of two sets.  I'd also have more room.  I could swap one of
the smaller drives to and have another backup of things like family pics
and such.  I haven't nailed down what I want to do yet.  I want to do it
right the first time.  ;-)  Oh, the second 16TB drive seems to have got
lost.  UPS tracking page says to file a claim.  Then it sent a text
update that it will be here tomorrow but doesn't show the box has moved
since it got lost.  I dunno when it will get here.  Maybe one of these
days. 

Right now I'm having discussions with that old 8TB SMR drive.  When I
try to update the backup data, it remounts read only and then I have to
unmount and remount.  I've ran fsck on it and it did do the usual fixes
but didn't report anything bad.  It also passes all the SMART tests,
short and long.  To test, I'm running defrag on the thing, it is at 20%
fragmentation so it couldn't hurt.  That should make it rewrite a fair
amount of data and see if it stays in write mode or goes to read only
again.  Doing this test on the NAS box in case it might be a cable or
machine problem, just to be sure. 

I think in the meantime, I'm going to avoid 20TB drives, unless it is
one heck of a deal.  I like the extra space but don't like odd log
messages. 

Now let us pray.  :/ 

Dale

:-)  :-) 

Reply via email to