NAS Compares Your WD Red NAS Hard Drives Might Be Using SMR – What You Need To Know

WD Red 2TB, 4TB and 4TB NAS Hard Drives Use SMR / DM-SMR

Over the last few days, a story has been circulating that revolves around WD Red NAS Hard Drives and their potential use of shingled magnetic recording (SMR) and a statement has been shared by the brand highlighting that this is indeed the case, namely, drive managed shingled magnetic recording (DM-SMR). Now, it turns out this is something that many platforms and solo users online had reached the conclusion of as early as July 2019 and when reaching out to WD, and was met with less than immediate responses. But now it appears that the WD Red 2TB, 4TB and 6TB Standard class NAS 3.5″ Hard Drives do indeed feature this method of platter utilization inside and it’s upset a few people indeed, but the details are not as black and white as some might think. So, what is SMR, Should you be concerned, What does this mean for bigger WD Red Capacities and How is WD responding to this? Let’s take a look at the news.

Continue reading...
- - -

Check out FREE NAS advice section on nascompares.com

 

[Coop777]

NOT good....what baffles me is the WD RED is marketed by WD as their NAS flagship drive and now we see yet another critical flaw.

Interesting to see further down the comparison to the HGST Ultrastar.

 

[Telos]

Uncertain which WD drives are SMR/CMR...

Read on...

On WD Red NAS Drives - Western Digital Corporate Blog

On WD Red NAS Drives - Data - Western Digital Corporate Blog

blog.westerndigital.com

This basically covers current production and not historic models.

 

[euri]

Hmm... A week ago one of my WD red 6T NAS drives (out of 4 identical drives altogether) went "degraded" after only two and a half years of use. Just by itself. No kicking, no pushing no touching, no performing a single routine with it, nor even looking at the unit. I 've already sent it back to SCAN computers for replacement. I wonder if this has anything to do with anything...

 

[jeyare]

I have read the blog twice:
- still is missing an explanation what the 180TB workload means in terms of device managed shingled magnetic recording (DMSMR)
- because the “DMSMR needs perform background data management tasks as needed” = additional workload generated by DMSMR itself. How much is it from the 100% base? Is it 50%? 60%? ... It’s really hidden or unclear information from WD.
- then you can consider what will happen (performance) when your workload meets this background disk management (managed by disk itself)

btw 180TB/y = 505GB/day = 42GB/hour (12h/day operation) = 718MB/minute:
- what is from basic point of view enough for home users majority. Really?
- when you go deep dive and your daily backup task contains more than 505GB (even valid for incremental backup) you are out of the target range and you have to purchase different disk model (brand)
- still missing how much from the defined workload will keep for the DMSMR.

I can see just 3 options, why they have such message for customers:
- they need hiding true story, because the story is even worse
- they have low cost PR team
- they don’t care about the 80% of consumers of Barracuda and Red disks, because such segment majority will purchase cheap disk anytime without a considering about consequences.

 

[WST16]

I’m not sure I read anything about when did WD made the change.
So old drives with capacities of 2TB-6TB could be CMR not SMR.

 

[jeyare]

I’m not sure I read anything about when did WD made the change.
So old drives with capacities of 2TB-6TB could be CMR not SMR.

but you have to take into consideration, that SMR is technology used since 2014

 

[WST16]

but you have to take into consideration, that SMR is technology used since 2014

I wasn’t aware. So all my 4TB drives could be SMRs then.

 

[Rusty]

I wasn’t aware. So all my 4TB drives could be SMRs then.

They probably are, yes

 

[jeyare]

another more important parameter which multiply your workload is a storage array usage (RAID, SHR)

then all the SMR active management processes (DMSMR) make impact to:
- disk latency - the time it takes for the sector of the disk being accessed to rotate into position under a read/write head. You can check it with command

ioping

- disk seek time. The time (in ms) it takes for the hard drive's read/write head to position itself over the track being read or written.
You can calculate subtotal of disk IOPS with formula:
IOPS = 1 / (Avg Latency + (Average from (seek time of write & seek time of read))
Note:
don't forget to normalize mili-seconds to seconds, because you will get pretty large IOPS

Another method for calculate current IOPS workloads for your disk(s) is based on RAID operation:
IOPS = READ I/O + (WRITE I/O x RAID penalty factor)
when:
READ I/O = Total workload x percentage share of read operation
WRITE I/O = Total workload x percentage share of write operation
RAID penalty factor for write I/O is:
RAID0 = 1
RAID 1(10 or SHR1) = 2
RAID 5 (SHR2) = 4
RAID 6 = 6

Finally you have to calculate your workload based on your disk array operation model. When you have RAID5 or SHR2 your workload must be multiplied by 4. Example from above:
Your daily expected write workload (max. expected) = 100GB = 8 533MB per hour (12h operation) = 711MB/minute = 12MB/second .... what is 6,7% of common HDD (WD Red) throughput, then check your average load.

Simplified calculation:
when your expected workload is (just for write operations) = 100GB x 4 (penalty factor) = 400GB x 365 days = 143TB/year .... seems to be near the threshold defined by 180TB. Uhh, and where is the read operation workload? Just use simple calculation for Read/Write ratio = 1:1, it depends on your operation model. And you are out of the range.

Clear calculation method:
you can use also converter from IOPS to data rate in MB/sec here
then when you have SHR2 based on 4kB block disk size, with:
107 IOPS for write ... your data rate is 37.9GB/day
133 IOPS for write ... your data rate is 47.0GB/day
your yearly workloads = (37.9 + (47*4)) x 365 / 1024 =80,5TB ... in "safe" range defined by WD for SMR
but when you change block size from 4KB to higher you are running out from the "safe" range


For measure of each mentioned steps you can use (ipkg) commands:
iotop, iostat, dstat, htop

Who need a help how to install ipkg, link is here

 

[jeyare]

this is analyze should confirm my feeling - they don't care about the majority of consumer's market
But I need to wrote:
- no one who use "j" class NAS is worried about performance. Except some power users here
- few percent from normal users thinking about the storage data architecture model, then there is a small group of users who calculate IOPS as one of disk (even NAS) choice reasons
- this is new area of knowledge for your future decision

PS: people are irrational and disk vendors know about it ... not just disk vendors

 

[WST16]

My plan was to get 2 X WD-10TB now, use them in the 216 and when we have the 918 replacement, get it and move the 10TB and the 4TB to it.
Now with this, I don’t think I should mix the CMRs and the SMRs.
Damn them!

 

[jeyare]

@WST16 try Seagate Ironwolf
To be sure no IronWolf Pro - they have SRM!

 

[WST16]

For some reason I‘m always attracted to WD. Never liked Seagate. I think I’ve had a bad experience with one in an external Lacie. Besides, I keep hearing people saying they’re noisy!

 

[Rusty]

I keep hearing people saying they’re noisy

The same goes for 4TB reds as well. In comparison, 3TB versions a much less noisy, but I'm in the same boat as you regarding WD vs Seagate

 

[jeyare]

Hmm... A week ago one of my WD red 6T NAS drives (out of 4 identical drives altogether) went "degraded" after only two and a half years of use. Just by itself. No kicking, no pushing no touching, no performing a single routine with it, nor even looking at the unit. I 've already sent it back to SCAN computers for replacement. I wonder if this has anything to do with anything...

Too many reasons possible w/o details

 

[chenks]

a lot of talk and stats etc etc.
but the bottom line to me is, what difference would it make to ME if my drives were CMR or SMR?

i bought 5 x 4TB seagate ironwolf, the article is about WD, but if one company is doing it then it would suggest others probably are too.

 

[Geeked]

I wonder if enough people complain about this will WD issue a recall? Oh boy I sure hope so because lying to people is never good. how else would a company "fix" this problem?

 

[jeyare]

I wonder if enough people complain about this will WD issue a recall? Oh boy I sure hope so because lying to people is never good. how else would a company "fix" this problem?

answer you can find in their blog - we don’t care

 

[jeyare]

a lot of talk and stats etc etc.
but the bottom line to me is, what difference would it make to ME if my drives were CMR or SMR?

i bought 5 x 4TB seagate ironwolf, the article is about WD, but if one company is doing it then it would suggest others probably are too.

then you are one of the lucky man