Who makes the best/most reliable HDDs these days?
>>106588109backblaze drive stats is your fren, anyways seagate is the worst don't buy it, for stuff like hgst you can even go for official refurbs
>>106588109They're all the same shit. Ignore the seagatespergs.
Anything that has a continuous operation rating and isn't Western Digital.
>>106588109It doesn't matter all that much these days. Find a good balance between price and warranty length. You should have backups and enough redundancy to survive the occasional drive failure. If you're lucky, it will even fail within the warranty period so you get a new one for free.
WD Red Plus or higher. Seagate Ironwolf or higher. Avoid all SMR drives e.g. Purples, Barracuda, Surveillance, Blue, Green etc and all 5400RPM drives.
>>106588109whatever happened to Maxtor?
>>106588410seagate
>>106588109Toshiba
>>106588109shitgate never again
How often should you preemptively replace your drives? Do you go by power on hours or general age?
>>106589646never, why would you?...you have multiple reliable backups of courseand if you don't: any drive can fail at any time so you are guaranteed to lose your data no matter what
>>106588109HGST designed drives, although they are manufactured by WD they are made in the old HGST factories.
>>106589899basically just ultrastar branded drives are using HGST tech and are made in old HGST factories. WD makes HGST designed drives that look similar to the ultrastar drives but are made in different factories (high end WD red for example).
>>106589646Every 17 minutes, give or take.
>>106589899>>106589920WD drives don't really exist.The last pure WD drive was like 14 years ago. There are a few WD designs still slumming around serving the low capacity drives (<6TB) but all drives 8TB and up are HGST designs.WD ceaded their own HDD development when they acquired HGST.
>>106589973WD still has the ultrastar brand for drivers that are made in ex-HGST factories. But yes, all modern WD drivers are HGST design and tech. The driver made in WD factories are less reliable than those made in ex-HGST factories.
>>106588109Definitely not Seagate
>>106589980They are all ultrastars in a sense.At the large datacenters, all WDs HDDs bear UltrastarOnes sold at retail as individual HDDs are sold as WD Gold but are the same exact HDDs.It's not indicative of where they are made, just branding.
>>106588335>Avoid [...] all 5400 RPM drivesWhy?
>>106588133I cant find any data on blackblaze suggesting or showing seagate skyhawk, ironwolf or x16 does any worse than other brands or models?
>>106590173because they're slow shit with high failure rates and most commonly SMR, why else?
>>106590277faster spin time, lower lifespan
>>106590297not with a shitty 5400RPM SMR drive kek
>>106588109>most reliablebuy an SSD
>>106588109Wd red plus crm 22tb drive
>>106588109They're all the same for your sample size of <20. Buy what you can afford to back up.
>>106589412>he does not have relevant informationTurns out the Seagate reliability issues were all counterfeit secondhand drives being sold as "new" or "refurbished" after being used for chinese chia cryptoscams.
Why not Seagate? I've owned a Seagate HDD and it has worked flawlessly for 3 years. It is the only external HDD I own.
>>106595209Seagate has an outsized share of lemons, but a 'lemon' HDD means like 5-10% failure rate for the model. Which is why I say everyone's sample size is so small it doesn't matter. If you buy 2 of the most unreliable drive on the market you still have an 81% chance of not experiencing issues.Conversely issues tend to occur in specific batches. If you happen to get a bad one, it's common to have 100% failure rate with a handful of matched drives from a model that's supposed to be statistically reliable.
>>106588109Barracuda 5400RPM SMR drives
>>106596566>SMRWhy?
The only drive I have that has failed over the last 20 years was seagate.
>>106589646When I run out of space.
is the russian schizo here? if not I summon him
is there really a big difference between samsung, wd, and seagate?surely they all fail at the same rate?
>>106589646If the sector reallocation count starts increasing rapidly or I start getting a bunch of read errors, I'll preventively replace them. Otherwise I usually only replace them when they abruptly fail.
>>106588133>backblaze drive stats is your fren, anyways seagate is the worst don't buy it, for stuff like hgst you can even go for official refurbsI won't own a seagate drive. Total shit. I do buy hgst/WD refurbs for RAID.
>>106589830>never, why would you?>...you have multiple reliable backups of course>and if you don't: any drive can fail at any time so you are guaranteed to lose your data no matter whatThis is the way.
>>106588109Toshiba, HGST/WD, Seagate in that order.
>>106590277Red Plus drives are not SMR and don't have high failure rates.Yeah they are slow but you can only get them up to 6TB so it's not a big deal *. Put them in a Z1 array and you get SSD level sequential speeds.* 8tb and higher Red Plus drives are just the HGST 7200rpm drives with firmware gimping to make them slow.
Is Toshiba good? I got their N300 8TB series. And thinking of getting another Toshiba 16TB to replace my Seagate NAS 8TB (has cartoons and tv shows)
>>106596625he wants to write at 2mb/s
>>106598361If you're buying Red Plus instead of Ironwolf Pro, you're an idiot
toshiba is worst and wd is best but it honestly doesnt matter at all the differences are minisculejust buy whatever has the longest warranty.>>106588133backblaze stats are next to useless for 99.99% of people
>>106598599>>106598361even at higher capacities, Seagate pricing and warranty dabs all over WD
>>106595127You're low information as well.When seagate first introduced their 12TB HDDs to DCs there where 3 separate issues that plagued a singular model of HDD.>platter contamination from production>premature head wear>contamination from the internal filtersIt got so bad Google demanded straight up compensation.Seagate 3TBs are also pretty famous for their failure rate.Seagate's bad reliability track record goes back way farther than Chia
>>106588133blackblaze are wd shills, I don't trust them. I'll buy exos and ironwolves all day everyday before I buy some overpriced wd red pos.
ssd are eternal
Any of you anons ever bought one of these renewed ultrastars?https://www.amazon.com/dp/B084X87F51
>>106595209Wow! Three whole years without catching on fire! That's a new record for seagate!
>>106599268Flash memory lifespan drastically decreases when more data is written to it and with high temperature. An SSD you write to once then store in your basement should easily last over 10 years but write a few dozen TB and keep it in a hot PC and it may only last 2 years. It actually gets so bad that once you hit the rated P/E cycle limit, they'll only hold data for a few days in warm conditions.
>>106598612Toshiba dabs them both.>>106598599>If you're buying Red Plus instead of Ironwolf Pro, you're an idiotIt's 10$ more for a half as loud drive. If I wanted to build a silent NAS (because I don't have a basement to store it in), that would be a good deal.
>>106599358Nearly any modern consumer SSD has so much write cycles that you'd take several decades to kill it. You'd need to be fucking NASA running 40+ year old satellites or 10+ year old mars rovers for this to become a real issue.If you you are seriously worried about this, check the SMART data on your SSD and divide the data written field with the power on hours to get the average written GB per day, then divide the TBW of the drive with that number. That will get you a ballpark of how long the drive will last with the amount of work you put on it. The last time I bothered checking that, it came out to some 80+ years and that was with a 860 EVO. Since then I moved on to Pro drives and a PM883, which have wear measured in petabytes.
>>106599268The moment 8tb ssds go below $200 the consumer hdd market is toast.
>>106599496>The moment 8tb ssds go below $200 the consumer hdd market is toast.ssd prices would need to halve at least twice, we are still almost 10 years away from that. And right now the only path forward for denser/cheaper SSDs is 3D stacking more and more NAND layers together, which will hit a hard gap with current lithography methods, and there are no new technologies to make them denser.Meanwhile HDDs are only just now ramping up heat/microwave/energy assisted writing production, 32TB drives are already out for enterprise buyers, and it is expected to scale up to at least 50TB, and bit-pattern/nanopattern platters are being researched which could double that density.By the time SSDs hit below 8TB, you'll be able to buy 32TB HDDs for the same price and average game install size for AAA titles like GTA 7 will have an install size close to a terabyte.
>>106599481Pic related is from a whitepaper written by a military hardware provider from 2020 investigating the effects of P/E cycles and temperature on flash longevity. Writing only once and leaving it at 75C barely gets you 10 years. I've seen cases where writing once and leaving it at 20C for a year will cause it to lose its charge. Don't fall for marketing speak "you're SSD is rated to write gazillions of petabytes*!"*based on our extrapolated data models in best case scenario
>>106599626Genuinely useful info. Thank you anon.From that we can extrapolate that SSDs are best used for frequently changing data or caching before using HDDs for long term storage.
>>106599646Whitepaper is from Curtiss Wright btw. This guy is also testing for himself how long some cheap SSDs will retain data. https://www.youtube.com/watch?v=igJK5YDb73w https://www.youtube.com/watch?v=rx3Y5x6uzKQYear 3 update should be up in a few months. Both the worn out and fresh SSDs are running into errors 2 years in. He leaves them in an unpowered state though. If they were powered on, the firmware would probably move data to healthier blocks before it becomes unreadable.
Amortize it based on the warranty period. All storage is doomed to fail. It's only a matter of budgeting and keeping backups.
>>106599626>>106599693Bruh you are speaking of two mostly unrelated thingsI'd almost consider it maliciousData retention is a different subject that write endurance.SSDs refresh their data over time so a drive left on has an indefinite lifetime only limited by write cycles.
>>106599716Data retention is most definitely related to write endurance. If I buy a storage medium, I want that device to hold my data. I'd buy RAM if I wanted it to disappear when I cut power.
>>106599735So I guess you never actually use your storage considering you will never power on your storage devices
>>106599741I just like my data to be secure if I leave it unplugged for more than a few days.
>>106599693dangis someone doing a test on "externally-powered" SSDs?thought they would've lasted at least a couple years just with constant access to charge
>>106599358>>106599626>ssds aren't good at decades of data retention in cold storage which means their lifespan is poor....these two posts literally have nothing to do with what the other two anons were arguing, an ssd that's powered up daily can in fact sustain petabytes of writes and last for decades, there's nothing wrong with such claims...the fact that data retention on a powered off ssd isn't as good as data retention on a mechanical drive has literally nothing to do with this discussion, why hdd fags always love to bring this up when nobody asked, no idea.
>>106599735>an ssd that can only retain data for a few years when powered off isn't a storage medium and pretty much as volatile as ram...the hdd cope here is absurd, long term storage is a niche use case that isn't what these drives are meant forme for example, I use my computer every day, rarely i'll be away for a few weeks at most and then come back, even if my drive were to be very worn and stored in a fucking freezer, retaining data for a couple of weeks is no issue, 99.9% of people don't just leave their ssds untouched for years with valuable data that's not backed up on another medium, it's literally a non issue and again, nothing to do with this discussion.you're somehow implying that bad data retention means bad lifespan, when lifespan merely affects the data retention span, it's the other way around and thus not really relevant to this discussion.
>>106599827Read the reply chain. I responded to >>106599268 who said >ssd are eternaland I provided evidence that they are not eternal. It sounds like you just want RAM but you're too poor to afford a mobo that has more slots so you use SSDs as cope.
>>106599604>By the time SSDs hit below 8TB, you'll be able to buy 32TB HDDsThe cheapest 8tb ssd costs $540 and they used to cost around $900 a year ago. In 2 years tops they'll be around $200. Meanwhile HDD prices have barely gone down, nowadays with around $200 you still only get 12-14tb.
Hitachi
>>106599866>the fact that long term data retention while powered off is bad means it's not eternalonce again, you somehow still believe these two things are the same when they're not.an ssd left unpowered might lose data, but it doesn't break, it doesn't stop workingwith "ssds are eternal" one is clearly implying eternal as in "it works" not as in "it retains data when left unpowered for long periods of time", there's literally no reason to think that guy was referring to cold storage retention rather than lifespan of the drive.>It sounds like you just want RAM but you're too poor to afford a mobo that has more slots so you use SSDs as cope....says the pajeet that will imply ssds are as volatile as ram to justify still running a 2.5" 5400 rpm hdd as his boot drive in 2025my fucking sides
>>106599626>Writing only once and leaving it at 75C barely gets you 10 years.Who here writes to their SSDs once at 75C and then lets them sit for 10 years?
>>106599914What part of STORAGE medium are you not understanding? Just because you never touch grass and you sit in front of a computer all day doesn't mean other people don't have lives where we might leave an SSD unplugged for a few weeks while on vacation. Nobody wants to babysit their tech just to have it function properly. And yes, SSDs do break. All that write leveling you're doing to it by leaving it on all the time can wear down the sectors and make them unusable. Flash storage is NOT eternal. It's a tradeoff between longevity and speed. Stop being autistic.
>>10659977210 years is significantly more than "a few days". You'd have to get something like a jail sentence for manslaughter to ever realistically come up with a scenario where you leave your SSDs unpowered for 10 years, and even then if you are not a retard you'd have a weekly offline HDD backup.Because SSDS ARE NOT OFFLINE BACKUPS. They are extremely high performance, always online storage. HDDs are for large density, offline storage (or always online in RAID parity arrays).
>>106599951The context of the graph is for choosing appropriate storage for military hardware in extreme conditions. It's an extreme example, but with other data it can be extrapolated to more common workloads. At the same temperature, an SSD that's reached 10% of its TBW threshold only retains data for 1 year compared to 10. It's something to consider if you value your data.As a side note, they actually found that writing data at high temperatures followed by storage at cool temperatures yielded greater data retention compared to writing at low temp.
>>106599869>In 2 years tops they'll be around $200.No, there was a large price dip in 2023 summer when everything fell to half the price, but that soon corrected itself. A SSD today is only a little cheaper than it was 5 years ago.8TB drives are a special case, they used to start far more expensive on account of their size and lack of competition, but that too corrected itself.From what I recall it takes 3-5 years for SSD prices to get hit significantly, we'll have to wait until 2026 or 2027 for the next major price drop like what happened in 2023. And by then you can easily buy 30TB HDDs for the same price.with that said if you look around on the used market you can get 8tb drives for 3-400$. I got 10tb nvme and a 8tb pm883 that way.
>>106599968>Because SSDS ARE NOT OFFLINE BACKUPS.Agreed, but this isn't common knowledge. Many people believe SSDs always outlast HDDs and think they're better for cold storage.
>>106599985>The context of the graph is for choosing appropriate storage for military hardware in extreme conditions.bro this is a consumer whoring board, nobody here runs military grade hardware in extreme conditions. even the most weel-to neckbeard on /g/ merely runs a giant fucking rack full with 4U servers and petabytes of anime.nobody here owns a missile control.
>>106599956You are the one being autistic.People who "touch grass" are literally the biggest adopters of SSDs, most won't ever see an HDD again. There are no masses of reddit threads or Facebook posts about how their PC won't turn on because they went on vacation. Not a single soul with a job or kids is worried about the data retention in their PC or god forbid MacBook.>>106600005People who think SSDs will outlast HDDs are actually keeping them powered on.
>>106600005>Agreed, but this isn't common knowledge.It's really common knowledge amongst anyone who follows computer hardware.>Many people believe SSDs always outlast HDDs and think they're better for cold storage.People who have that horrible misconceptions deserve to get their data lost. I bet they also still burn shit on DVDRs.
>>106599869>In 2 years tops they'll be around $200.If anything they will be back to $900 with AI garbage driving up demand for data storage.
you're better off getting a kioxia or sas ssd. any datacenter ssd really if you want best bang for buck.
So, on this ssd data retention thing.Do ssds automatically resilver themselves if left powered on or something?
>>106590173HDD binning also involves testing drives for the amount of vibration, so 5400RPM models are considered low tier.
>>106588109Probably best to use disks from multiple manufacturers. That way if there's a major defect with one brand/model you're not completely fucked.
doesn't matter, i just get whatever old SAS drives have a good deal at the time.
>>106601244This hasn't been true for over a decade now, since they stopped making 5400rpm drives in general.
