Who Makes the Best Hard Disk Drives? 444
Hamsterdan writes "Backblaze, the cloud backup company who open sourced their Storage Pod a few years ago, is now providing information on drive failure rates. They currently have over 27,000 consumer grade drives spinning in Backblaze storage pods. There are over 12,000 drives each from Seagate and Hitachi, and close to 3,000 from Western Digital (plus a too-small-for-statistical-reporting smattering of Toshiba and Samsung drives). One cool thing: Backblaze buys drives the way you and I do: they get the cheapest consumer-grade drives that will work. Their workload is almost one hundred percent write. Because they spread the incoming writes over several drives, their workload isn't overly performance intensive, either. Their results: Hitachi has the lowest overall failure rate (3.1% over three years). Western Digital has a slightly higher rate (5.2%), but the drives that fail tend to do so very early. Seagate drives fail much more often — 26.5% are dead by the three-year mark."
That's interesting (Score:5, Informative)
Re:And what about... (Score:5, Informative)
Re:Ignorant to their own research (Score:5, Informative)
They're looking for 4TB models. They only cite two models without any further information.
Seagate ST4000DM000
vs
Hitachi HDS5C4040ALE630
You can look up technical details, benchmarks, etc. but perhaps the decision is simply in the price.
Seagate: $164.99
Hitachi: $295.00
For the Hitachi model to start making sense, price-wise, that Seagate model would have to fail a lot more than their numbers are currently showing,
( And yes, I'd imagine they can squeeze better deals than regular consumer prices out of the companies - but then, they could do that for either brand, and probably through an intermediary anyway. )
Re:Amazing how times change. (Score:5, Informative)
Are you sure about 5TB drives being available on Amazon for a couple hundred dollars? I didn't think Western Digital had released those yet and their website shows 4TB as the max capacity for their Green, Black and Red series drives and Amazon doesn't have any listings for 5TB drives? If they are available can you share a link?
Re:Ignorant to their own research (Score:5, Informative)
After all this research, Backblaze still pick the highest failing drive.
Disclaimer: I work at Backblaze. Every month we ask a list of about 20 suppliers for their best price on a variety of drives. There is a little spreadsheet we have that kicks out which drive to purchase based on those prices and drive failure rates. Even if Hitachi is the very highest reliability in our application, it only justifies a SMALL price premium because when one drive dies, we don't lose any customer data. It saves our datacenter IT team 15 minutes to *NOT* swap a drive, so that's worth 15 minutes of salary to us, but not more.
Depends on model (Score:5, Informative)
So it's more nuanced than "Seagate stinks, Hitachi rules." (Hitachi is a subsidiary of WD now, operating separately only because that was a condition China placed on them before they'd OK the merger.)
Re:Sad to hear (Score:1, Informative)
You do know China is communist, right?
Re:Amazing how times change. (Score:2, Informative)
"Have to agree. We used to love SeaGate and despise WD - but now you can get 5 TB WD drives on NewEgg for a couple hundred bucks."
Here's the thing, though: as they say, they buy "the cheapest" they can. WD has very clearly defined categories of drives, and they don't even try to claim that "the cheapest" are by any means either the most performant or the most reliable. If you want those, you have to get the yellow (enterprise) or black drives.
Of course, other manufacturers might have categories too, but Seagate doesn't really... it just has regular and "enterprise".
So it's not surprising that the cheapest WD drives have the rates they do... because they're the least reliable drives they try to make.
Re:That's interesting (Score:3, Informative)
20x12x11=2640, with an implied average use time of 5.5 years. Five or less failures means a MTTF of 2904 years, an average failure rate of 0.035%/year. This is a large enough sample size to have statistical significance.
But you're obviously to lazy and arrogant to run the numbers.
Re:Amazing how times change. (Score:5, Informative)
NewEgg doesn't sell any 5TB drives, internal or external, and nobody makes them either (yet), so... no you can't?
Re:Amazing how times change. (Score:4, Informative)
Was this a Seagate scandal or actually a MiniScribe scandal (acquired by Maxtor, acquired by Seagate)?
Re:More Backblaze slashvertising (Score:4, Informative)
Actually if you look they do call out model numbers and even talk about which ones they won't touch and I find this VERY interesting and informative! As it happens my 30TB of space happens to contain a mix of both "reliable" and "unreliable" drives according to their testing. I run a mix of sizes from 1.5TB to 3TB using unRAID and as drives fill up they get upgraded. I have a few 1.5s that they call out as being trash (ST31500341AS) and an EARS drive of that size that should probably go ASAP since they are well into their second if not third year of use. I actually happen to be running a parity check right now and once I got past the 1.5 drives speeds increased a great deal. Once past the 2TB models things got even better so the 3TB drives appear to be much better performers. Naturally they list the ST3000DM001 as having a 10% failure rate too so I'm not exactly doing handstands! The replacement drives are all that model and I've been playing with them in another system to try and come up with something better for my needs than unRAID and so far nothing has come out much better so into the array I guess they will go here shortly.
My hat's off to Backblaze for publishing this and letting consumers know who's got decent drives and putting feet to the fire those that don't!
Re:Amazing how times change. (Score:5, Informative)
RAID-5 on a system where you don't think you have high quality drives, a high quality power supply, battery backup (for the RAID card) plus a high quality UPS unit (preferably multiple UPS units hooked up to a set of redundant PSUs inside your case) -- is simply a bad idea. Sooner or later, you *will* lose the array due to a double-drive failure. Oh and make sure that you have plans to swap out drives on a regular basis and have a working backup plan.
RAID-6 is better, but not by much. It can at least deal with a double drive failure. But performance still goes in the gutter while it's degraded and/or rebuilding.
One of the more fault-tolerant setups is a 3-way RAID-1 mirror where you can lose 2 of 3 drives without losing data. The downside is that it is only 33% efficient while RAID-6 (1 spare, 2 parity, 5 data) is 62% efficient. A well configured RAID-10 setup also works well, but never gets much above 40-45% space efficiency if you set aside a hot spare for it.
Main reason why I prefer RAID-10 for larger arrays is that the time to rebuild a failed disk is linear to the size of a single disk within the array (because you have mirror pairs). With RAID-5 / RAID-6 the rebuild time scales with the total size of the array. For a 15-20 drive array, that means RAID-10 could rebuild the failed drive in 1/5 to 1/10 the time of the RAID-5 or RAID-6 array with the same number of spindles.
Re:Amazing how times change. (Score:4, Informative)
You might want to read their report as they point out that swapping drives is COSTLY. Their experience is that it's better to pay a little more and not have to screw with the drive. Their report also documents which drives "pop out" of their RAID arrays and require costly attention. When a RAID array goes "bad" it can take time to recover, that's a cost that is almost certainly going to be more than what that troublesome drive saved them in the short run. These drives don't cost much more than $100 apiece and I'm betting their employees aren't being paid minimum wage so that hypothetical $50 savings isn't much especially if data is lost....
Re:Amazing how times *don't* change. (Score:5, Informative)
And in the last 25 years I've replaced about 200 failed drives. The Maxtors failed at about 5x the rate of the Seagates. That shows you want ancidotal evidence is worth.
To know actual reliability, you need stats on the level of Google or Amazon, that can tally failures by the 10's of thousands.
Re:Ignorant to their own research (Score:3, Informative)
I'll get a longer list, but we buy from places like B&H Photo, NewEgg, Amazon, Central Computer (it's a retail outfit in the San Francisco Area). Some months we still "drive farm" where individual employees show up at Costco or Best Buy and empty the shelves "retail". At only 700-ish 4 TByte drives per month, Backblaze is not allowed to contact the drive manufacturers directly. We have been told if we purchase 10,000 drives in one purchase then we can deal directly with Seagate or Western Digital.
To pre-empt the next question, the reason we don't just buy 12 months supply and stock pile the drives is because the expectation is that drives drop in price every month. Stock piling 12 months of hard drives means over-paying for drives "in advance". Of course, this ended up being a mistake during the Thailand drive crisis (doh!) We're still trying to figure all this out, your opinion is every bit as valid as ours!!!
Google has already done this (Score:5, Informative)
Re:Amazing how times change. (Score:5, Informative)
I have RAID 5 with 4 x 3TB seagate drives. 1 drive failed after a year and the 2nd in the same NAS failed a couple of days later before a replacement could come in the post. So far 3 / 8 Barracuda drives failed in just over 1 year. After just losing 4TB of data, including my entire photo collection, I've sadly realised RAID 5 isn't enough.
Phillip.