Hey can someone recommend me a HDD anywhere following these criteria’s
Shipped from EU
8-12 TB
good quality
budget 1220 SEK
I wanna get links to items nothing else
thats what i look for
1 Like
I am a big fan of the WD Ultrastar DC HC530 14TB SATA 6Gb/s 3.5-Inch Data Center HDD - WUH721414ALE604 0F31152
One word of caution, it works great in a server case, but if you try to use it with a typical PC PSU you will also need to buy a power adapter (by using a simple 4-pin Molex-to-SATA adapter to supply power to the HDD) or something to disable power to pin 3.
1 Like
In my opinion, hard drives today are commodities. Brands don’t matter, various marketing gimmicks they introduce to somewhat set themselves apart are from insignificant to meaningless. All that matters is cost per TB, including power consumption over its useful life. That’s it. That’s the only metric.
(This is of course assuming the drive is suitable by all other factors, i.e. not SMR, with ERC adequately configured, etc).
1 Like
Cars are commodities as well, but there is a real difference between models. The Western Digital UltraStar HDDs (as well as Seagate EXOS) are data center class drives with longer life and lower power costs. Cost is not the only factor, often when you pay less you get less.
Well, hard drives are not cars. They have one job: keep data for specified period of time at minimum cost. How they accomplish that — is irrelevant.
“Datacenter class drives” is meaningless marketing mumbo jumbo.
Decision on whether it’s worth buying these drives only depends on one variable: (upfront cost/expected lifetime + average power*power cost). That’s it. If it happens to be “most best” by this criteria — great.
But if, say, prosumer Seagate IronWolf suddenly is on sale or you found them used very cheap and that brings this parameter below that of “datacenter class drives” - then that’s the better drive at the moment.
“Longer life” is also voodoo magic. All drives eventually fail. Drives are designed to be used in redundand arrays, so failure rate is irrelevant. If one model has 1% failure rate (on anverage, I.e. not applicable to your specific workflow) and the other is 100 better — it’s still less than 1%. Irrelevant, as it only contributes as much to cost.
1 Like
I would add warranty. There is no reason to accept anything below 5 years if buying new HDDs.
Well it MATTERS if the company makes quality products.
SEAGATE didn’t at some point, and the stench has followed them to this day.
IT Matters if company is cheating customers (like kingston did with unfamous replacement for lower quality parts without telling anybody)
It matters if a company is trying hard and has good values, vs a company that is pure cash grab scum, like M$, or like SEAGATE.
From perspective of Storj Labs, it doesn’t matter.
But as a SNO i need best HDDs at lowest prices.(not running any array and never will!;P)
Coz im not gonna replace it every 1-2 years, thx but no. not at this payout rate.
(“… ironwolf” - OMG, pls don’t even say that loud!)
(I need to fill one node before I buy another HDD… I need to fill one node before I buy another HDD… I need to fill one node before I buy another HDD… I need to fill one node before I buy another HDD… I need to fill one node before I buy another HDD…)
Sometime when you see storage deals, it can be hard 
Well, this is a general statement, but it is not applicable here: the industry figured out long time ago that doubling the quality (e.g. halving the failure rate) is more than twice more expensive, exponentially so. Put it differently, every dollar put into improving MTBF brings logarithmically diminishing returns.
Put yet another way – nobody will pay double for halving the already low enough failure rate
So the concept of RAID emerged: Redundant Array of Inexpensive Disks. (Now it was retroactively changed to “Independent Devices”). This allows to drastically improve failure rate at slight increase in cost. It’s like finding a cheat code.
Since this point, as long was the failure rate of an individual device is tolerable, its a bad idea to keep trying to improve it, for everyone involved.
Specifics for skeptics
Let’s look at worst case numbers. Assume we have 5 drives, $100 each, with 1% annualized failure rate, and we won’t even consider bad sectors, we’ll assume the disk completely fails.
So, to lose data, we need any of the drives to fail. This has probability of data loss 1 - (1 - 0.01)5 = about 5% in the first year.
Now consider 6-disk singe-disk fault tolerant array, that provides the same amount of space as 5 loose drives. To lose data within a fist year, we need to have one disk fail, and another disk fail within the same week (amount of time it may take you to replace the first one): 6 * 0.01 * ( 7 / 365 )2 = 0.0022%
Thus, we get 450x better “quality” at the 20% increase in cost (extra drive to provide fault tolerance).
Warranty is meaningless. I would completely ignore it.
If the probability of annualized drive failure is 1%, then 5 year warranty value is 5% of drive cost. I would not split hairs about 5%. You shall also subtract cost of shipping the failed disk back to manufacturer, which, depending on where you live, can be 5-10% of cost of disk, entirely negating all savings due to warranty Buying used drives on the other hand gets you 50-60% of value. 10x (or infinitely, if you consider cost of shipping bad drive back) better.
If you expect to replace every disk every 2 years, it means it has 50% annualized failure rate. These disks don’t exist 
1 Like
Amazon has free shipping for renewed drives if they fail and prices are mostly same all the time
no i mean like having 10-20 of these ironbluff, and every 2 years or so 1 can start looking like dying, f that. I have like 80GB HGST disk made in 2010 or so, in operation, and they are like “aaarggg… its haaard but i can go! its fine!”
This is too small of a sample size to draw conclusions about the whole operation.
I had opposite experience, which is also anecdotal and just as irrelevant: every single ironwolf still spinning, but all HGST drives I ever had died on my watch. They are also usually more expensive so I recently happen to not buy them. (I don’t look at brands, they just never fit my search criteria which is price per TB). Recently it’s all sorts of OEM branded disks, Dell EMC, IBM, etc (which are Segeate in disguise of course).
I think right now I only have one ironwolf in NVR, 4TB, spinning since 2019-20. Does not seem to want to die any time soon. Those irovolves, 4TB ones, specifically, were absolutely magnificent – lowest power consumption I have recently seen from any drive.
well there’s must been some sort of … worse and better sorts (like those big global food brands, ex: redbull etc, there was some scandals, they produced better quality for germany, and separate lower for eastern european countries, must have been something similar with the drives lol
You guys in USA somehow got diluted HDDs lol
well sure if they spin 5200 rpm only on paper and in reality probably half that lol
no srsly, i hate seagate. sorry.
Well, taking into account:
We can say, HGST and WD perform best. Although, annualized failure rate of almost all hard disks is below 2% making 5-years survival >90%. Implying that better survival at maximum may cost 11% more. And usually, WD and HGST cost much more than 11% above the lowest competitor.
Bigger 12+ TB drives seem to survive better than 8-10TB, although not that much different to bother.
Since power usage usually isn’t more for bigger than for smaller drives, this is virtually always benefiting bigger drives. Power usage usually differs only 1 or 2W (=9-18kWh/year). So the difference actually isn’t here either.
So, essentially, statistically the best thing you can do is buying the biggest drive with lowest $/TB.
2 Likes
Me too, but not that much to see that their price given previous view point is worth it. Because 30% lower disk price is really worth it.
Bought a renewed HGST drive off swedish amazon. 1500 SEK with case. HDD is 4.2 years with good health. My new node i set up 24h ago is working well
1 Like