HDD & SSD Durability in 2023 - We Need to Talk

[@djashley2002]

I sometimes get the feeling that durability of HDDs is the luck of the draw. I've had disks fail catastrophically within three months of purchase, and I've still got disks that are close to twenty years old in daily use and still function perfectly.

[@kazeee22]

20 years???? That's insane what brand?

[@DevilbyMoonlight]

Very true.. I remember the 1GB palladium drives which only lasted a few months , some seagate 1.5 TB drives fell over but I still have a 20mb MiniScribe HDD out of an Amiga 2000 and it still works!

[@aiqo4413]

probably linked to their transportation ?

[@blunderingfool]

The SSD that came out of an Acer Aspire x-something or other still goes to this day... and the first thing that happened to it was my sister absolutely belting my brand new computer off the table the same day we bought it, because elbows. =P@@kazeee22

[@modbox9560]

Eventually SSDs will fail before mechanical the caps on nand flash which holds your data when drive is unplugged without caps SSDs would be RAM which loses data every power cycle another reason RAM has lifetime warranty

[@marklewus5468]

About cold storage: most SSDs do not refresh individual cells unless being written or read. If using them for cold archival storage, over time the charge on the gate of each cell’s gate leaks off and that cell may become unreadable. This is exacerbated for TLC and especially QLC. If you use a SSD for cold storage, the best approach is to format it and completely rewrite it at least once a year. OTOH hard drives maintain their magnetic state almost indefinitely so the issue there is primarily how long will the bearings and capacitors last. I have many hard drives that are over 20 years old and still 100% readable.

[@syarifairlangga4608]

do you think expensive ssd automatically refresh the charge of their cell

[@DevilbyMoonlight]

Before the internet was widespread I ran a BBS for 7 years on an Amiga, the drives were running 24/7 all year round no failures.. the drives are small by modern standards the largest being a 2GB SCSI - with SSD's I can give you a real world case a vertex3 drive bought when they 1st came out is still in use, its gone through several machines in the last 12 years and has been living in the Mrs laptop for the last 3 years, shes a very heavy user of way over 12 hours a day - and its still going strong

[@barygol]

I still have my vertex 2 running

[@skepticalmechanic]

I remember the BBS days… it was magic!

[@bjarne431]

I think for the vast majority of users SSDs are by far better, if only high capacity SSDs were cheaper…

[@JohnSmithWesson]

reading people's reviews about their HDD purchases scares me, many reviews where drives have defects, some doesn't even work, some work but there is already some errors, some works fine but fails after few months, i think most of such problems is because of how HDDs being transported by retail companies and delivery workers, most people does not acknowledge how fragile HDDs are, that's why many consumers get defective HDDs

[@mrguiltyfool]

Hdd is still king in terms of storage per dollar. I love ssd but there is no way that i can afford 4x 12tb ssd atm

[@skepticalmechanic]

Only get scared when a Dr tells you cancer is going to take your life soon…

[@chandanankush]

Any thought on keeping SSD always on and HDD always on, if we ignore read write considering super low usage. Which one is better ?

[@mk72v2oq]

HDD wears constantly while turned on. And if it constantly spins down on inactivity it is far worse. SSD does not wear by simply being online and read operations are effectively free.

[@mk72v2oq]

@@Prostreetsburnoutparadise yeah, HDDs are immortal, cool story bro. And stack of dead HDDs on my shelf is a hallucination. Also check your eyes, I said SSDs do not wear on READ. Of course they wear on write, everyone knows that. They do not wear when turned off. They could lose information after a very long time offline (years, theoretically, there are no real reports of that happening), but it does not affect physical health of the disk itself. Although, HDDs indeed have a good use as a cold offline storage (with proper storage conditions). But if you actually read the original comment, the question was about "always on" scenario.

[@mk72v2oq]

@@Prostreetsburnoutparadise HDD does wear meaningfully during operation because of its mechanical nature. Spindle and heads failures are a thing even in perfect working conditions, you can't deny that. Modern HDDs tend to become less durable because of density increase and overall complexity. Especially SMR ones are a joke. Yes, magnetic surface itself can last pretty much indefinitely, but it doesn't really matter if HDD died for whatever reason, you can't reuse them. Also the surface could be damaged by failed mechanical part, it is not so uncommon. SSD memory does not wear on read operations, don't talk bs please. Electronic components (most often a controller) can die pretty much any time, but it is a shared problem between HDDs and SSDs, it does not depend on the disk wearing level.

[@greenzero3389]

In 2016 I bought 4 x 3TB WD Reds which lived in a QNAP 421 in Raid 5. Used in a home environment so not a great deal of use. Last month one failed and then a second and third failed during rebuild. So I now have 3 large paperweights and 1 untrusted drive. I did make the mistake of buying all the drives at the same time from the same etailer. I've replaed them with 2x WD Red and 2 x WD Blue drives in two RAID 1 volumes.

[@benyomovod6904]

I always buy the same size but different manufacturers hdd

[@Raintiger88]

I have only one NAS with HDDs remaining, all others are SSD. I just built a new desktop - 22TB of SSD storage. Time is the enemy of HDD, writing (especially write amplification) is the enemy of SSD. I have personal experience with an unpowered raid 5 all SSD setup for 3.5 years and it was perfectly OK. I read an article once from Toshiba that claimed 10 years, but I'm not sure I'd put that to the test for important data. I have several old NAND devices that have gone much longer (18 years). For writes AT A MINIMUM, QLC is 1000 PEC, TLC is 3000 PEC, and MLC is 10000 (note that there are variations with newer NAND tech). Even if I were using an all TLC SSD nas for something write intensive, I wouldn't be worried. I've purposely tried to kill a few NAND drives by doing as many DWs as the write speed would allow all day, every day. No failures, but my sample size was admittedly small. I'm hoping that I'll eventually see someone else's testing for something similar.

[@SnaxMuppet]

It would be great to have just the summary of what all that meant to me as a home user... it was all a bit technical tbh for me to make anything of it.

[@bru2al1tyusa82]

Thanks for making a video on this subject maybe there is a way to go in depth on home backup drive recommendations

[@YozoDerzu]

I've had mine WD Elements 500GB HDD since 1 oct 2015 and it never failed on me. I was surprised it was this old actually. Good stuff. So I bought another new one 1TB frm WD Elements and I hope it will run just as good.

[@The_Ballo]

Did they mention bit rot while unpowered or did I miss it? SSDs have a miserable shelf life

[@essdee800]

Watching now. Thanks friends.

[@christerwiberg1]

I am a bit curious, I have a Synology DS918+ with 2x 8 TB-drives, that has been running for almost 4 years by now. I soon have to expand, but I am thinking that should I just buy one more 8 TB drive and run for a couple more years or should I buy for instance 2-3 x 12TB new drives? Guess I can depend on my existing WD Red Pro drives some more years (and Yes, I have backups). Any thoughts about that?

[@IntoxicatedVortex]

Typically I would say go with bigger drives. You need to plan out your cycling of old drives that have been running for many years (I look to do it at between 3 and 5 years - less for 2.5" HDDs). If you're adding bigger drives, by the time you're cycling out your oldest drive(s) you can then also raise the total pool size if you're using RAID (SHR is simpler but slower). Thing is, what is your NAS replacement cycle? When factoring this as well it is often more cost effective you simply fill the bays as cheap as you can to then keep any savings for the next NAS. When the new NAS comes with best value drives at that time then you can re-task your old NAS as a hot replicant (via Synology Drive Sync or other) and/or as a backup. At this point you also gain yourself a redundancy for the NAS. There will always be a moment that the NAS itself will need to be factored into the costs.

[@christerwiberg1]

@@IntoxicatedVortex I do agree, seems reasonable, getting one more disk, aiming for an upgrade of the NAS in a year or two having my current as a backup. Then I can cancel my Backblaze account if I have that as offsite backup. Would be an idea 😀

[@scoria1755]

I can't believe what anybody says about their HDD and SDD because I can't verify their UPS.

[@Br0adCastYourS3lf]

Why is the backdrop of NASCompares mirrored?

[@gem2148]

What is best for long term cold storage?

[@alexjerry-yy3sb]

hitachi

[@bjarne431]

For cold storage mechanical disks are safer, but if you do power your SSDs, i think, at least some days each year (they need to refresh the stored data) then they are arguably better (moving components in mechanical hdds are scary to trust long term)

[@TheTrueCyberpig]

I have a 920+ running 3x6TB drives and want more. I use SHR for home use and media. Need more space. Suggestions. 1x20TB? Would that work or would that screw the SHR current setup. Confused. I will be.

[@torstenhelgo1485]

There is still one Question no one could answer me until now: How does turning the drives on and off affect their lifespan over time? Some people say it is beneficial because it saves power and gives the hard drives some time to rest. Others say that the damage done by accelerating them from 0 to over 5000 rpm is so big, that it shortens the lifespan to an extent, where it does not justify the saves in electricity. So, what is it? Does Synology’s power-saving Mode make sense? It shuts down the drives after 20min if they are not used, just to boot them back up 20-30min later, because some syncing process begun. That constantly turns the HDDs on and off. Should I disable this power-saving mode?

[@ottogarza]

Very courous about this same, i was thinking if give the nas a time like 8am to 8pm is usually when i work and then put the disc to sleep will be helpfull for the disks or not, i mainly use as backup for video so they are not in constant use because i dont even edit from the nas

[@mk72v2oq]

Excessive parking wears HDDs a lot. I mean, if you turn it on and off like once a day there is nothing to worry about. But spinning it down on inactivity periods is actively harmful.

[@edwarddejong8025]

We bought top of the line enterprise mechanical drives 9 years ago and they have held up fabulously well, only 1 or 2 failures out of 50, and run them 24/7. SSD's do wear out, and that is one of the only remaining advantages of mechanicals.

[@bjarne431]

And mechanical hdds does not wear out? I disagree, its just less predictable then SSDs :-)

[@ponderoso2009]

No disrespect, but II can see why Eddie is in the toolshop 😂 Yes please, more hdd sdd reliability vids - you possibly need to find half a dozen benchmarks/data points you can more or less evenly compare the drives and come up with your own NAS Compares' score... maybe it'll catch on and manufacturers will follow... then again 😅

[@QuoVadistis]

David Bellamy lives! Hoorah...

[@EthelbertCoyote]

Great discussion as a 3D generalist and video creator I changed my normal drive setup this go round and bought a Optane 905p drive for my swap files drive for windows adobe and a few other files simply to save drive wrtites and reads for my other drives life. As the Optanes durrability is insane and it's low q depth io is also hopefully if will be a winner for that workflow. I do have a question however for a main video working drive does it make sense to raid 0 (with back up of course) that drive to reduce reads and writes and hopefull reduce temps? I have been using gen 3 nvme drives till now and have not had to worry about heat and then throtteling would raid help here or not? Thinking with Raid 0 read/writes should be haved per drive yes/no?

[@lukastemberger]

I don't think trying to build a system that will last 10 years is reasonable. Drives will keep getting faster, larger, more reliable and cheaper. I don't want to be stuck with old tech in 10 years. Even in 5 years...

[@guyb7005]

wish they could dumb this down for general public

[@rodrigofilho1996]

The problem with SSDs is that they die without notice, smart is fine one day, the SSD dies the next day, HDDs usually give notice way before the die, they will make noises, smart will give errors, with hard drives u can see they are dying then save the data on them.

[@fnorgen]

I've had trouble with an HDD corrupting data at least half a year before showing any other warning signs. in this period it ruined my OS image multiple times, before performance started noticeably degrading and I realized what the problem was. To be fair though, I didn't lose any valuable data. It was just really annoying! I've also got an HDD that I've kept running in my computer since 2007. It's dreadfully slow by modern standards topping out at 70 MB/s, but it just keeps happily chugging along, and I've gotten really curious of just how long it can last.

[@redslate]

Sure, but you don't know what or how much data is going to be lost in between your HDD sending you those mixed signals. 😅 This is such a backwards argument... An SSD is going to last a full decade longer than a comparable HDD. Data redundancy is the only proven safeguard.

[@tombouie]

Hi Eddie but ?where was the sea-gull though? ;)

[@nascompares]

As I read this message, I am trying to record a video special on the History of Kickstarter NAS...so far filming has been delayed as (what sounds like) 7-8 seagulls are either humping or killing each other above my roof.... Don't push me Tom... Don't push me!

[@tombouie]

@@nascompares Hilarious ;)

[@nascompares]

I'll be honest bud..you are kinda, sort IN that video now, because I forgot to turn off notifications

[@JohnSmithWesson]

I would go for SSD and archive importand data on optical discs in multiple copies

[@redslate]

Why optical?

[@JohnSmithWesson]

@@redslate in my experience they're pretty good for long term storage. I have CDs and DVDs from mid 2000s. It is still better to overwrite the same data on new disks with at least 2 copies as a precautionary measure.

[@redslate]

@@JohnSmithWesson I mean, I've got CDs from the 90s that are still nominal, but I wouldn't trust them for important documents. Solid State is the way to go. If you need a cost-effective, hi-cap medium, magnetic tape has been the defacto means for the better part of a century.

[@JohnSmithWesson]

@@redslate if something happens to SSD then there is no way you will ever recover data from it, with HDD it's possible.

[@redslate]

@@JohnSmithWesson That's a played-out, rationalized argument. If you send your HDD to a forensics specialist and pay hundreds of dollars, _maybe_ you can recover most of your data (you still won't get everything). Otherwise you're stuck with whatever you manage to pull from disk yourself *after* you've already begun to lose valuable data. An SSD simply won't fail for several times the life of a HDD, so you'll have already migrated your data well before hand. Also, modern SSDs are designed to fail-open, so you can still read/copy the data without specialized assistance.

[@andybb]

My Seagate 18TB Failed in my NAS last week went straight on their website and was still under warranty got the new one this morning with great service all the way from Holland as well

[@YozoDerzu]

Seagate is low tier quslity. Buy either Samsung or Western Digital. Preferable WD. I have my 500GiB HDD for 8 years now. Zero failures whatsoever.

[@josephp1592]

@@YozoDerzu LOL. I have a 10+ year old seagate that still works, what is your point?

[@redslate]

Seagate has some of the worst AFRs and MTBFs in the industry.

[@AchmadBadra]

​@@josephp1592 seagate 2014+ are trash.

[@NicolasSilvaVasault]

@@YozoDerzu but everyone says wd is the worst of the bunch? who the hell tells the truth here

[@AinzOoalG0wn]

bought like 4-5 Seagate ironwolf non pro 4tb. 1 had bad sectors 2 years into it's 5 year warranty. the rest which i got at the same time still seem to be ok. rma process seemed to be easy. there was even an easy collection and also shipping back to me for free. Seagate seem to have a good rma at least in 2023. Not sure about WD. HDDs eventually fail.... and you may or may not need to RMA while your drive is still under warranty, usually 5 years at least for the ironwolf series which is commonly used for NAS.

[@OrlandoPaco]

Those numbers should be seen as base numbers only! Nothing is de-facto!

[@nascompares]

Of course... But also, even if nothing is defacto, industrial production standards have to exist regardless. Else how else can one choose a drive over another in terms of durability.

[@OrlandoPaco]

@@nascompares Buy the one's that have those numbers really high as in work time, on time and number of drives, ultimately hoping for the best.

[@GQNetwork]

Thank you for the great comparisons. I am currently reviewing to purchase a configuration with a NAS Synology RS2423RP+ with a Synology E10G21-F2 Dual Port 10G SFP+ PCIe 3.0 Ethernet Adapter (wish this was standard) along with twelve Samsung 8TB 870 QVO 2.5" SATA III Internal SSDs. This will be running on a 10GbE home network. I decided to add the SFP+ card to account for heat generation and performance using a RJ45 to SFP+ 10GbE transceiver. I also decided to configure using the SSD cards because of the noise level and heat that HDDs generate. Otherwise, I was planning to go with the WD Red Pro 18TB HDDs. The rack system is in a data closet that happens to be next to a media room and a few bedrooms. This system will be used for video editing, raw photos, data file storage, documents, and remote data access through the Synology VPN configuration. Your advice and guidance on this system will be much appreciated before this investment is implemented. Thank you for another amazing lesson on network system configurations!!

[@IntoxicatedVortex]

Personally I would maybe look at doing a mixed 2 pool implementation where your hot storage is SSD and your warm storage being HDD. This enables you to get the best of both where your active data is on SSD and then your archive/backup etc is on much larger HDD. For instance 5 x 4TB SSD in RAID 5 plus 7 x 18TB HDD in RAID 5/6. You can then easily move data between the pools via File Station internally. Noise levels for HDD aren't a big issue given you've already accepted going with a RackStation, otherwise, look at the DS3622xs+ instead. By the way, I would not recommend the Samsung QVO in RAID. Just don't do it. Go with the EVO at a minimum else better the WD Red else best the Seagate Ironwolf SSDs. While you get access to 8TB with the QVO the positives with it quickly stop there.

[@pzwolski]

I would not bother with WD REDs anymore when there a GOLDs a little bit more expensive. A small difference in price, but huge in specs.

[@GQNetwork]

@@IntoxicatedVortex Thank you for the recommendation. That is truly a great setup that I will explore. Thank you.

[@GQNetwork]

@@pzwolski I will definitely look into the Golds as I am not familiar with their specs. Thank you.

[@alvinnorin8820]

Calling 544mb/s "insane" really made me check the age of the video lmao.

[@nascompares]

In context! vs HDDs in a NAS.

[@WR3ND]

Let me know when someone makes a 16TB MLC SATA SSD at a reasonable price, and we'll talk. But no, seriously, that's what I'd like.

[@WR3ND]

I'll take 5, please.

[@redslate]

It's existed in enterprise for a couple years now (unless you mean Double-Layer MLC). It's still too cost-prohibitive for consumers at this point.

[@WR3ND]

I did mean double layer. I have some 4TB MLC SATA drives. 5 16TB drives would be nice for a NAS RAID, I'm thinking, but preferably they would be very reliable. Cheers.

[@mk72v2oq]

It won't happen. Modern 3D TLC is good enough. Despite popular beliefs, SSDs most often die because of other reasons like a controller failure far earlier than the memory cells have a chance to actually wear out.

[@WR3ND]

@@mk72v2oq I wouldn't trust it in a RAID as a NAS.

[@tofu_golem]

I just had an M.2 cache on my Synology go belly up.

[@BMSMusic1]

My Seagate barracuda 160gb last for 8 year becaming fail to read and bad blocks eaten fast in her last lifetime,500gb seagate pipeline as second drive last about 3 year and die bios cant detect it and i replace with extrenal hdd seagate 1 TB for backup and rarerly opening it just fine until now,and i replace die 160gb system with wd blue 250g it only took 1 years and she die 🤣 not detected in bios its my first awful experience with storage,later on i buy and switch back to seagate barracuda 250g run fine and after 3 years using that switch to SSD 250gb for system the old just for backup os if ssd became failing and i easy cloning the os in hdd to new ssd....

[@user-wl7dt1uw2e]

Not sure you were able to deliver on this. I didn't really get a sense here. How about this thought? Most consumers are scared S less about drive failure and don't push the lifespan and simply replace drives. They never run them to point of failure due to fear. So do those drives get destroyed ultimately? What I'm saying is that there is such a small sample size, I don't think you can find enough data to be conclusive at all. Just fear and paranoia. Look at WD. Oh, warranty over. WARNING. Imagine if consumers EOL everything once warranties expire. If you can find a business that runs their drives to the point of failure, then use them for meaningful data. Most businesses get to the 5 year warranty and replace. Now the question is do those go back into use elsewhere. They won't if influencers want to say that 5 years is a great time to replace a WD Red for example. You also make a great point that when people write to a drive, that data stays there. Nobody wipes data. We place it, we store it and we add to it. That's it. So does Blackblaze duplicate this usage? If not, let's discuss that very point!!!!