Which Enterprise Class HDD? Or not necessary?

[Evertb1]

though I don’t expect I’ll ever need the middle one.

Some toolless solutions for (hot)swap bays and the like, do depend on all the holes. For the synoloyg caddy I needed to cut the middle pins. Doable but not to my liking.

 

[Evertb1]

What drive do you think this is? He or Air?

They were coded 120EMFZ. It's not straight forward what they are. Some think they are actually WD140EFFX drives that are firmware-locked to 12 TB, others think that it is a firmware-locked 14TB Ultrastar DC HC350. So are they He? I am not sure,

 

[MalVeauX]

Hrm,

Prices are up everywhere on externals compared to before. Maybe the market pulling on them and the obvious shucking videos and tutorials are going to change this. Seems to have already changed. And now its just a matter of time.

So the real question is, what is the white label drive now in there? Is it an ok drive (probably is, but what is it)? And does it compare to another drive really?

So for $213 you can get on Amazon right now a 12TB Elements external. Whatever white label is inside.
Or, you can get a SeaGate 12TB Compute Pro standalone drive for $244.

The rest seems to jump significantly after that for "NAS" labels and other color or purpose labels.

So maybe the white label drive in these externals, being only a bit cheaper, are the only non-purpose drives out there right now.

Seems like there's a big gap going from 8TB to anything else currently.

Very best,

 

[elorimer]

I think it might be that high platter count drives don't have room for the middle screws on the bottom. Anyway, I noticed that on the 8TB EMAZ drives I got. My Antec case from years ago has caddies that used those screws (2 of the 4 mounting points.)

The bigger surprise is that these are not 5400 rpm drives. They are 7200 rpm drives that have all the power, the noise, and the heat, but I guess failed QC at 7200 so they are called "5400 class" drives and have the SMART code forced to 5400.

 

[Evertb1]

Until now I only shucked drives for non-critical applications. The 12TB drives from my friend are the only ones I ever shucked for main storage. It's not that I didn't trust the drives in an USB enclosure. It's just that I like to know what I have. And until this moment I was not able to find any conclusive information about the specifications of those things. My friend should have looked at this before he bought them.

 

[elorimer]

Even in that video the guy refers to WDRed specs (i.e., 5400 rpm).

 

[Evertb1]

Even in that video the guy refers to WDRed specs (i.e., 5400 rpm).

Search on google for 120EMFZ. See what conclusive information you get. I could find nothing official from WD. Just the general disclamer about disks in external enclosures. But I found a lot of guessing on all kind of sites and forums. I guess they are 5400 but is that because they are 7200 drives forced to 5400 by the firmware? Or are they really WD reds for example? You tell me.

 

[elorimer]

I think they are 7200 drives spinning at 7200 but the data throughput isn't up to spec, so WD has put 5400 in the rotation rate field so you aren't expecting 7200 throughput performance. It may be the firmware slows the throughput performance, but not the spindle speed. But I agree, I'm just speculating without any official info.

Because 5400 is in the rotation rate field, my 8TB EZAZ drives are routinely identified as 5400 rpm drives: Ebay, Walmart, Amazon.

 

[Constantin]

Yup, as one OEM I encountered put it: I am free to sell an appliance at a lower efficiency than it actually operates at. In this case, it was two appliances, identical to the damn last screw whose rated efficiency and price point differed.

No different with these hard drives - the audio-spectrum analysis by some reddit users confirms the rotational speed beyond a doubt - dominant 120Hz frequency = 7,200 RPM drive, while 90Hz = 5,400 RPM drive. The statement given to blocks and files from WD only underscored that truth further. Sadly, I wouldn't expect anything less from WD these days. They're trying to milk that cash cow as hard as they can. Being in a 3-way oligopoly helps!

 

[Evertb1]

The statement given to blocks and files from WD only underscored that truth further. Sadly, I wouldn't expect anything less from WD these days. They're trying to milk that cash cow as hard as they can. Being in a 3-way oligopoly helps!

Yes, it's a sad world where some smart chosen, legally correct sentences are of more value to a company then the authenticity of the products they produce. Talking about customer satisfaction.

 

[MalVeauX]

Hrm,

Well, I went for the gamble. The WD Easystore 14TB drives are $229 (61GB/$). I'll update with a report when they arrive of what is inside the enclosures. Looks like it will take a week to arrive, so maybe these will be "newer" than some old-stock drives and I'll see what WD is up to these days.

Too bad we cannot just tell a drive what speed to run. That would be interesting.

Very best,

 

[Evertb1]

The WD Easystore 14TB drives are $229 (61GB/$)

Well wish you the best of luck. I have just seen the arrival of the first batch of WD Red plus marked drives at my favorite store. The 14TB cost eur. 429.00. A 14TB WD elements would set me back around eur. 300,00. Easystore's are not sold in the Netherlands even not in the nl store of Amazon. Of course I could buy them at Amazon.com: $ 500,89 (price + shipping). Not today thank you

 

[MalVeauX]

I was going to just go with 12TB for now, but just thinking of ZFS in general, I figured, if I'm going to put 9TB on a 12TB disc, it's already going to be likely too full on capacity to continue adding anything or much more, to allow plenty of extra room on the disc(s) for ZFS to function ideally. So being a $10 difference between the 12TB and 14TB I went ahead on the 14TB to get that extra 2TB for ZFS basically but I have no plans to fill up the 14TB drive(s), just to have the additional space for ZFS processes. After these drives are in and happy, I will add a 8TB mirror pool for two separate mirrors and that should be it for a while for my purposes (I already have a 8TB ultrastar so will add another and that will be that).

Very best,

 

[MalVeauX]

Update:

I received the drives today. These were both WD 14TB Easystore externals from BestBuy directly. I plugged the drives in via their USB and powered just to see if they worked and see any info on them. Then used that information to look up what these drives might be. The drives are identical, and both show made in March/April of this year. Here's the info:

Model: WD140EMFZ-11A0WA0

Found this info and it matches this it seems:

Reddit - Dive into anything

www.reddit.com

Here's the CrystalDisk info on each:

And after I verified they worked, I shucked them and it was pretty easy to do it without any damage with some thick cards. I put them into my FreeNAS server and booted up. They did not spin up on their own with my EVGA BQ PSU, so the 3.3v thing matters on these. I already had a molex -> sata adapter, so I used that, and they spun up powered just fine. So that part still applies. They are missing their middle screw hole on the sides as I found in other threads, so I had to do a slight modification to my HDD cage holding bits but that's simple and doesn't matter much.

FreeNAS happily saw them, used them and I built my first pool with them. I started copying data and using them to see how things behave and will keep monitoring how they act for a while.

Very best,

 

[Evertb1]

Found this info and it matches this it seems:

So if that information checks (and it looks to be from a very informed source) those are 7200 RPM drives, while the information in the rotation rate field shows 5400 RPM. And WD is getting away with it just because they sell them in an USB enclosure. But there are industries where that kind of shenanigans would be considered to be bordering on fraud. But anyway I hope that you enjoy them.

 

[MalVeauX]

So if that information checks (and it looks to be from a very informed source) those are 7200 RPM drives, while the information in the rotation rate field shows 5400 RPM. And WD is getting away with it just because they sell them in an USB enclosure. But there are industries where that kind of shenanigans would be considered to be bordering on fraud. But anyway I hope that you enjoy them.

It just goes back to what I was wondering about white labels just being lower binned drives that didn't make the cut for higher tier drive quality/consistency and so they get firmware stamped lower and put into an enclosure and sold elsewhere to avoid loss. I'm not certain of this, but I'm just guessing that this is a possibility.

Very best,

 

[MalVeauX]

I was trying to read more about the maximum capacity that is suggested to safely use ZFS without it going into space optimization vs performance. I'm less worried about performance and more concerned with just having data integrity. So is 80% still the suggested maximum usage one should do across the total capacity of a pool? I was reading that 90~95% is where it starts to shift to a different optimization scheme. I'm at 31% used capacity with this pool currently and will put another 5TB of data onto it by tomorrow, leaving around 3TB of space free (75% capacity will be used, so I can go a little father maybe, but I don't want to push it, I'd rather the drives and ZFS have plenty of extra space to focus on integrity).

So far I've put 20ish hours on these drives with non-stop copying of data. So far, no errors, no memory errors, no read/write/heal errors reported anywhere. Everything sounds good and the temps are staying 33C at all times. Speeds are good enough for my purposes so far.

Very best,

 

[jgreco]

It's more complicated than just picking some percentage. It really has to do with the kind of data you're sticking on the pool, and if there are block rewrites or small files or other fragmentation-ish things.

For example, if you're storing VM virtual disks, you probably don't want to get beyond maybe 30-40% pool capacity. As you stress out ZFS by reducing the ease with which it can find free space, performance suffers. This is just a natural side effect of a CoW filesystem. How bad the impact is and whether it bothers you are different issues.

If you are creating an archival system where you write once to the pool and then never delete or update anything, you can run up to 98% and probably never notice much performance degradation, because ZFS isn't struggling to find contiguous space.

Those probably represent opposite extremes. The 80% seems to work out for many people, but doesn't for others.

 

[MalVeauX]

It's more complicated than just picking some percentage. It really has to do with the kind of data you're sticking on the pool, and if there are block rewrites or small files or other fragmentation-ish things.

For example, if you're storing VM virtual disks, you probably don't want to get beyond maybe 30-40% pool capacity. As you stress out ZFS by reducing the ease with which it can find free space, performance suffers. This is just a natural side effect of a CoW filesystem. How bad the impact is and whether it bothers you are different issues.

If you are creating an archival system where you write once to the pool and then never delete or update anything, you can run up to 98% and probably never notice much performance degradation, because ZFS isn't struggling to find contiguous space.

Those probably represent opposite extremes. The 80% seems to work out for many people, but doesn't for others.

Thanks,

The data is mostly videos, music and photos, stuff that doesn't get updated really, just read over the network and served out. Makes a lot of sense with VM's that are constantly updating, growing, changing. I have a few backups for our phones and our computers and stuff but they're small amounts and it's mostly additive backup, rather than full overhaul each time.

So sounds like I'm ok to go higher on capacity use and be ok then. This is just one pool of two, so I don't plan going to something like 98% on one pool. Was just curious more about where these numbers comes from. It makes a lot of sense with the two examples you provided being extreme and why.

Very best,

 

[joeschmuck]

So if that information checks (and it looks to be from a very informed source) those are 7200 RPM drives, while the information in the rotation rate field shows 5400 RPM.

Hum... I didn't see that anywhere, the drive model WD140EMFZ-11A0WA0 does not match the linked article model WUH721414ALE6L4.

Lets talk about how hard drive physics... The heads need to fly on an air bearing (or He) in order to not destroy the head and the platter surface. That air bearing is determined by the rotation of the platters, also called the Bernoulli Effect (fluid dynamics). The heads must now exert enough downward force towards the platter surface to ensure there is a specific gap between the heads and that platter surface, a.k.a. Head Flying Height and is typically less than 6 nano-meters apart. To far away and the magnetic field is too weak to write/read. Too close and the head hits the platter causing physical damage to both head and platter, what we call a head crash.

So a drive designed to spin at 7200 RPM will have a head with a stronger downward force than a drive designed to spin at 5400 RPM. So if a drive that was deigned for 7200 RPM operation were to be slowed down to 5400 RPM, and the heads were not changed out for 5400 RPM specific heads, the drive would be highly susceptible to head crashes. I don't believe that using He vice Air would change this difference other than it would take less force to push the heads down with He to the proper distance than with Air. 'He' only provides less resistance to spin the platters and allows 7X the heat transfer, so you can pack more platters into the same space and thus have a larger capacity hard drive.

I've been working on hard drives since they were 14" oxide disks and the heads were moved with a hydraulic piston. Basically it was a custom IBM 1311 drive that used a custom 1316 disk pack. The custom part was the disk pack was doubled with 12 platters (20 usable surfaces) at about 9" in height and the drive was a ruggedized version that could be used in military applications, and the electronics had to be migrated from common circuit boards into thousands of replaceable circuit cards that were heavily shielded.