Which Enterprise Class HDD? Or not necessary?

[MalVeauX]

For 8TB, yes, see separate thread. Just shucked another three this summer.

For 10TB I have reports of Red Plus Air and He, with Air more recent.

For 12TB I don't know. This is why CrystalDiskInfo first, followed by one shuck of a drive, followed by careful identification (you'll see visually whether He or Air and the ID of the drive is still on the label, separate from the part number), followed by shucking "all of them" if previous steps were satisfactory.

Thanks; interesting. I'm thinking I'm going to focus on the 12TB~14TB options for now. The price per GB is essentially the same across the Element drives (single drive versions) from 8TB to 14TB. I was noticing that the dual drive Elemental enclosures advertise they have Red drives inside, but the price is terrible on those.

Very best,

 

[Constantin]

FWIW, I went with refurbished 7,200 RPM helium HGSTs for my server and 5,400 RPM shucked helium drives for my backups. I had a few early failures with the refurbs (that's what burn-in is for) and the rest have worked just fine since. None of my shucked drives have failed so far but they tend to be used on a very intermittent basis, so the comparison is not fair.

The sweet spot re: GB/$ keeps moving but its also important to point out how OEMs are directly responsible for the value-pricing going on re: server-grade vs. shuck-grade hard drives. The oligopolistic HDD industry seems to have convinced itself that this is the best way to milk the HDD cash cow as the all-flash future creeps closer. Like DRAM and other supplier industries in the past, the current HDD industry seems like a juicy target for an anti-trust collusion investigation.

 

[MalVeauX]

FWIW, I went with refurbished 7,200 RPM helium HGSTs for my server and 5,400 RPM shucked helium drives for my backups. I had a few early failures with the refurbs (that's what burn-in is for) and the rest have worked just fine since. None of my shucked drives have failed so far but they tend to be used on a very intermittent basis, so the comparison is not fair.

The sweet spot re: GB/$ keeps moving but its also important to point out how OEMs are directly responsible for the value-pricing going on re: server-grade vs. shuck-grade hard drives. The industry seems to have convinced itself that this is the best way to milk the HDD cash cow as the all-flash future creeps closer.

That seems to be the case; the shuckable externals definitely have the price advantage. They're making money either way. So everyone wins there. I may try a shuckable pair of 14TB drives and check what they are before opening them and report back. They're sold by Amazon with their no-questions-asked return policy. The current $/GB ratio is pretty much the same across the externals here regardless of size after 8TB. Everything else is significantly lower ratio.

$520 for 1:1 mirror redundant 14TB basically.

++++++++++++++++++++++++++
++++++++++++++++++++++++++

I have a few old drives in service still that I just checked with CrystalDiskInfo:

Hitatchi 500Gb 7200RPM drive (HDS721050CLA362) has Power On Count of 1507 and Power On Hours of 66,699 hours. 7.6 years.
WD 1TB 5400PRM drive (an old Green version, WD10EADS-00L5B1); Power On Count 1129 and Power On Hours of 66,082 hours. 7.5 years.

Good heavens. I expected the Hitachi to still be going strong. But wouldn't think this old WD Green drive would still be doing a good job. Yikes.

I have newer drives running but they're young compared to these, mostly WD Red 8TB and WD Ultrastar 8TB, but only 2~3 years on these and I would expect them to at least get to this point. Never expected a lousy green drive to make it 7 years though.

Yea, I need new drives... hah. Time to get these 14TB Elements and shuck then.

Very best,

 

[Constantin]

I'd wager that HDD life is a probability function with a few external factors thrown in for fun.

• Backblaze's use case shows how some drive families behave better over time for their use case.
• Backblaze's drives run at 17-30*C on average, i.e. lower than most SOHO applications. Drives running hotter than this tend to fail quicker. Mine stay between 28-35*C, with the upper limit only getting hit after multi-day, heavy operations.
• Stable power, ESD protection, 5,400 RPM spindle-speed, etc. likely also contribute to long, trouble-free life.
• Conversely, frequent spin-ups (for power savings), heavy I/O, higher-spindle speeds (7,200RPM+), etc. likely shorten life.
• Helium may help reduce power needs in the short term at the cost of making the drive un-repairable /un-recoverable in the long term.
• Firmware issues may also influence life

Most of these factors are un-knowable to the average consumer up front. All we can do is review the data out there, see what use cases are similar to ours and make our purchase decision from there. Sometimes, it works out and sometimes it doesn't. Once the market gets wise to opportunities (genuine WD Helium CMR Reds in EasyStores!) the OEMs make changes to keep the artificial delta between their value-priced HDD tiers.

 

[MalVeauX]

I'd wager that HDD life is a probability function with a few external factors thrown in for fun.

• Backblaze's use case shows how some drive families behave better over time for their use case.
• Backblaze's drives run at 17-30*C on average, i.e. lower than most SOHO applications. Drives running hotter than this tend to fail quicker. Mine stay between 28-35*C, with the upper limit only getting hit after multi-day, heavy operations.
• Stable power, ESD protection, 5,400 RPM spindle-speed, etc. likely also contribute to long, trouble-free life.
• Conversely, frequent spin-ups (for power savings), heavy I/O, higher-spindle speeds (7,200RPM+), etc. likely shorten life.
• Helium may help reduce power needs in the short term at the cost of making the drive un-repairable /un-recoverable in the long term.
• Firmware issues may also influence life

Most of these factors are un-knowable to the average consumer up front. All we can do is review the data out there, see what use cases are similar to ours and make our purchase decision from there. Sometimes, it works out and sometimes it doesn't. Once the market gets wise to opportunities (genuine WD Helium CMR Reds in EasyStores!) the OEMs make changes to keep the artificial delta between their value-priced HDD tiers.

Good points; plus outliers. Every system will have a low end drive that performs like a champ, and high end drives that are lemons. I definitely want to gamble the in-between though. I don't need massive throughput and speed, no major I/O, just a large file sever for media and to server as a backup to my other system(s). The critical stuff gets burned to M-Disc Bluray 100Gb optical discs as my 3rd (or 4th in this case) physical copy of anything I want critical redundancy on.

I wouldn't expect WD to keep semi-premium drives in a shuckable external, especially with so many tutorials on YouTube basically telling you to do it. It's not like how things were back in the 90's for sure, in terms of how fast information spread. These days you can watch a video tutorial on how to do anything the day before it's even an option. Hah. But perhaps desktop class drives are adequate here, for cost, with ZFS and redundancy. Just looking for 12TB desktop class drives by WD, there are not really much, compared to others (like the Seagate Baracuda Compute). Looks like almost everyone beyond 4TB is moving towards these mass capacity drives and instead of trying to keep desktop class drives around, like WD Blacks and Blues, they're all moving to the SSD game where they're up to 2TB already. I just hope the shuckable white labels are more like archival class drives and less like the old green drives.

Very best,

 

[jgreco]

12TB WD Easystore (Shuck) ($218)

Well, no, what you have to do is wait for it to go on sale. And I don't mean the weedy $219 sale price, I mean the $179 or $189 that pops up for just a day or two every so often, like about a week ago when they were $189.

 

[MalVeauX]

Well, no, what you have to do is wait for it to go on sale. And I don't mean the weedy $219 sale price, I mean the $179 or $189 that pops up for just a day or two every so often, like about a week ago when they were $189.

I will watch for sales daily and strike next time it drops.

Though I may also watch the 14TB version ($259) to drop for a day and get those, to ensure a bit more overhead for ZFS capacity.

My current data that I want to migrate onto this redundant pool is a combined total of about 9TB. So a 12TB would already be 75% filled. 64% if I used 14TB. I realize this will not be my only pool for long. I will have two pools. This first pool make it easier to migrate the second. So this first pool could be 12TB or 14TB and the next pool will likely be 8TB (x2, mirrored). That should handle my next 2~3 years of data generation after culling. I generate about 1TB of data per year after culling on average so far. I generate 2~4 TB of data actively but then cull back anything not needed a few months after creation and my trend is growing but it's not growing exponential, so I'm comfortable in the 1TB per year rate with room to spare. The second pool will buy me 3~4 years at this rate. So in reality probably 3 years.

Very best,

 

[Yorick]

as the all-flash future creeps closer.

I don't see that ever happening, simply because of raw materials. There's not enough available per annum to service the storage needs of the industry with just flash. I expect we'll see flash and HDD alongside each other for the foreseeable future. If His Muskness ever gets asteroid mining to be economical, that might change. I don't expect to see that in my life time.

 

[MalVeauX]

I don't see that ever happening, simply because of raw materials. There's not enough available per annum to service the storage needs of the industry with just flash. I expect we'll see flash and HDD alongside each other for the foreseeable future. If His Muskness ever gets asteroid mining to be economical, that might change. I don't expect to see that in my life time.

Until the flash is in a medium or container that is extremely resistant to magnetic and radiation effects, we will likely need other things.

I would love for the future to use dense matrix solid material and etching; imagine lab grown crystal in layers and the ability to write via engraving/etching and read via laser. Sure, write once only likely. But the read-only application for archival purposes would be amazingly resilient and likely outlive our species. Maybe find itself in the hull of a spaceship where radiation is a big concern (where flash will only survive in hearty resistant materials).

Very best,

 

[MalVeauX]

Hrm,

Come to think of it, Labor Day is coming next week. There's usually sales for any particular calendar event these days. I'll watch a few items and go with what has the best cost per Gb.

12TB or 14TB WD External (Easystore, Elements, etc)
12TB or 14TB EXOS SeaGate

The externals will likely win. But you never know.

Very best,

 

[Constantin]

I don't see that ever happening, simply because of raw materials. There's not enough available per annum to service the storage needs of the industry with just flash. I expect we'll see flash and HDD alongside each other for the foreseeable future. If His Muskness ever gets asteroid mining to be economical, that might change. I don't expect to see that in my life time.

To some extent this is an artifice of the market. There appears to be far more competition on the SSD side than on the HDD side. I won't discount that the DRAM chips may be coming from a limited set of foundries, however.

For example, the other day I had to choose between a used 2TB enterprise NVME module and a "new" 2TB Toshiba CMR 2.5" drive. The cost delta was so low, I went for the NVME drive. Had I chosen a consumer-grade NVME module (WD Blue), it would have been close to parity, even if the NVME module was new. Now a SMR drive would have been a lot cheaper but its performance would have potentially been terrible, so no thank you.

The price trends for all things microelectronics continues unabated, but not so much for HDDs, or they would be cheaper than they currently are. The periodic flushes of cheap HDD storage seems to signal that HDDs could be sold at significantly lower price points but that the oligopoly is keeping prices high in general and dropping them only occasionally to opportunistically skim the price-conscious end of the market.

Put another way: Only in the HDD market is it cheaper to buy an enclosure with a drive than just a bare drive. That in itself speaks volumes about how upside down the HDD business is. However, I have no doubt that the HDD OEMs figured out that this was the way to maximize profits, just as other industries have (change, seat assignment, baggage, time-of-day, route, etc. fees in the airline business).

I doubt it has anything to do with asteroids, however, as there is plenty of silicon here on earth (though we have occasional shortages of pure silicon production capacity).

 

[elorimer]

I send them down to the electronics shop, where half a dozen guitar picks pull open the shell without any trace of damage, put the parts in a baggie, which goes back inside the shell, which goes back into the box, which then goes into storage for two years, just in case a drive fails - because the externals have a 2 year warranty.

Exactly, and most credit cards extend the warranty by a year.

 

[Evertb1]

In my opinion, you want actual legitimate guitar picks to open these enclosures.

That will work for sure. I am a guitarplayer (somewhat) and the picks I have will do the job. But when I needed to shuck a couple of Seagates I just cut some wedges from a couple of old plastic cards like credit cards, membership cards etc. After rounding of the edges I turned them into usable tools. Just don't use the thin ones. They are to weak.

 

[jgreco]

That will work for sure. I am a guitarplayer (somewhat) and the picks I have will do the job. But when I needed to chuck a couple of Seagates I just cut some wedges from a couple of old plastic cards like credit cards, membership cards etc. After rounding of the edges I turned them into usable tools. Just don't use the thin ones. They are to weak.

Exactly. The thing that the guitar picks really have going are the beautifully bullnosed edges.

You definitely do not want to use a metal tool with any kind of sharp edge. If you jam any sort of metal tool with a hard edge in there, like an iSesamo, you can easily leave notable scratches, and even with nonmetallic spudgers, some are made of very hard materials which can damage the case. And then of course if you jam something thick in there, you will break the plastic fingers/clips on the case, which also just screams that someone was tinkering.

For a successful opening, you need to pry all of the clips simultaneously just a bit, and that's really the win for guitar picks, because they're small, strong, with a bullnosed edge that makes it easy to get in there.

 

[Evertb1]

Some of my high tech shucking tools. Cost a bundle though but worth the investment. And yes, I can pick strings with them as well. But I must admit that the products of Fender do a better job of that . Can't help it, Fender for me all the way.

 

[Stilez]

I'm using seagate for HDD, Intel and Samsung for SSD. At a pinch HGST or whatever they are now, but the price has never been worth it. There used to be reports about seagate and reliability, those were specific models many years ago, ignore them. What I like about seagate is, I've had around 100-200 drives and they last ages (In use currently includes 2 x ST31000524NS Constellation 1TB's from 2010, a Barracuda 7200.8 from 2005, and a ton of older 6TBs). They're fast. And the warranty replacement, the few times Ive needed it, doesnt care if I bought them new or 2nd hand, just wants to know what date.

When you pick a supplier for disks, which can fail, check out their reputation for warranty ease. You'll need it some day.

 

[Evertb1]

This weekend I shucked 2 12TB WD elements for a friend and I discovered that shucking is becoming an adventure these days. They were already tested in their USB enclosere (took ages) so I could start with shucking them. Out of their enclosure I tested them in my lab rig to see if they showed up. They didn't because of the 3.3 v. 3 pin (a known problem). testing with a molex sata adapter to get done with that problem worked just fine.

That 3.3 voltage problem can be solved several ways. Isolate that pin on the drive (piece of tape for example) ore use a molex sata adapter on the power side. I don't like the risk of a piece of tape on the loose in a slot so that one was out. And a adapter isn't a possibility in every situation. Luckily the drives were destinated for a Synlogy NAS and the recent models can handle that 3.3 v. problem just fine.

And then I discovered that WD has added another problem. They have made changes to the threaded moutingholes of the drives. I wish I had taken some pictures but I try to describe the problem.

Normal 3.5" drives have three threaded mounting holes on each side. On these drives the middle threaded holes are not there. The bottem of a drive has 2 x 2 threaded holes. On these drives two of those threaded holes are moved to the corners of the drive. So if you are depending on those for the retention of those drives you are out of luck or must improvise something.

The synology caddy use plastic retention strips on the site of the drive. Those strips have three pins that click in the holes of the drive. We just removed the middle pins of those strips and we were good to go (though mechanically a bit less strong). But I am pretty sure it is not always that easy.

 

[Yorick]

3.3V workarounds are typical for Enterprise drives put into consumer gear. I modify the cable on the PSU side, since the 3.3V pin serves no purpose unless the system understands to use it for spin-up control.

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

Good to know on the mounting holes, though I don’t expect I’ll ever need the middle one.

 

[Constantin]

I would not be surprised if OEMs start to ship drives with custom features for the external market that make them unfit for use in the internal market just so that the "bare drive rents" for drives destined to be used in a NAS can continue to be collected. At the scale that these guys are building stuff, the marginal cost of having two different form factors is almost nil but the profit motive is definitely there. Plus, no post-sale consumer can complain about the contents of an external drive "not fitting".

If this becomes a more wide-spread trend, expect more creative / flexible mounting solutions in the aftermarket.

 

[MalVeauX]

And then I discovered that WD has added another problem. They have made changes to the moutingholes of the drives. I wish I had taken some pictures but I try to describe the problem.

Normal 3.5" drives have three mounting holes on each side. On these drives the middle holes are not there. The bottem of a drive has 2 x 2 holes. On these drives two of those holes are moved to the corners of the drive. So if you are depending on those holes for the retention of those drives you are out of luck or must improvise something.

That's interesting, also not great news, but maybe not that big of a deal depending on the enclosure. As long as some of the mounting holes are still going to fit most standard stuff out there, it still should be doable and the worst thing would be possibly just "pin removal" on the enclosure side.

Thanks for the heads up!

Very best,