Failure Rate Analysis - Best 10Tb+ hard drives: Seagate, Western Digital or Toshiba?

I perform an in depth AFR (annualized failure rate) analysis on 10Tb, 12Tb, 14Tb and 16Tb hard disks using 230,000 drives SMART data as a data source. We find out which manufacturers perform best, and which models are the lemons to avoid. All these vendors state their drives have an AFR of 0.35%, but who is really giving the accurate picture?
Video on the broader analysis of 430k drives over 10 years of data : • Comparing Seagate vs W...
Enterprise vs NAS disks? Which should you choose? : • Exos vs IronWolf Pro -...
Link to the BackBlaze data source : www.backblaze.com/cloud-stora...
You can support me at: www.buymeacoffee.com/sometechguy
Thank you to everyone for watching and hope you enjoy the content!

[@PrincipalAudio]

If it's a big enough sample to provide a chart that shows high enough resolution data, it's big enough for me to make a decision off. If it's a sample size of, say, 5 disks, then the resolution just won't be there. I'd say the data you're presenting here is more than enough for people to make a valid decision on. Thanks so much for posting it and the hard work you've done.

[@sometechguy]

Appreciate the feedback and comment. 😊

[@dquiznoes]

The amount data presented to make these conclusions is great and very in-depth! thank you for your work!

[@sometechguy]

Thank you, I appreciate this! The data from Backblaze is fantastic and contains a lot of detail and a huge volume of stats, its over 410 million rows, each with around 35 columns of good usable data, with many more that isn't so useful. They also did a great job in keeping the data pretty clean, I worked with really dirty datasets before that require a great deal of clean up but the BB data is highly consistent. But it still takes a lot of work to extract the trend data and aggregate it up so you can get the AFR stats. Glad you found it useful.

[@Pulverrostmannen]

I have a Seagate Hard drive that been running for more than 139000 Hours, that's almost 16 years of 24/7 and it is still going, it just recently re-mapped for the first time 8 sectors to spare ones but so far nothing else gone. it have load/unload the heads 5 million times so far

[@sometechguy]

That's impressive, which model is it? 😎

[@Pulverrostmannen]

@@sometechguy ST96812AS 60GB 2,5” I actually think when it comes to 2,5” drives Seagate been the generally best. And WD when it comes to 3,5”. Worst ever is Hitatchi Travelstar for 2,5. And the IBM Deathstar lol. I had very bad luck with drives in general but I also seen very long lived ones in my days too

[@Helios.vfx.]

Hi, how do you remap the drive?

[@Pulverrostmannen]

@@Helios.vfx. Typically a normal drive will automatically re-map sectors when needed by itself through its built in Smart function. You can monitor this using software such as Gsmart control and Crystaldiskinfo. You can manually do a sector scan of a drive and search for bad sectors using Gsmart control or Windows/programs to check the drive for problems before they eventually get big enough to cause major problems. When a drive is getting worn out and it re-map a lot of sectors or mark many as problematic it is a sign of eminent failure and you should replace the drive. For example: reallocated sector count. This is the value of successful remaps that the hard drive done for you both on mechanical and SSDs. A value higher than 0 is a sign of age or increasing risk of failure. Current pending sector count, this value indicates the drive have problems with sectors but it is waiting for a suitable time to re-map it or check again. It may also be that it is failing to do a re-map of these sectors hence it is a pending problematic sector. Increasing values here have a big chance of damaged data as data is often still stored in these inaccessible sectors. Uncorrectable sector count. A value here indicates the drive have problems with sectors and could not correct them with the built in smart. This is often caused by locked data due to file systems on the drive. Your data is at eminent risk of failure as your drive cannot correct damaged sectors. If your drive have all the 3 combined in higher values you can be assured you have to backup immediately because your data is at maximum risk of complete failure. Monitoring a drives Smart is a very good thing to do at all times as all hard drives are constantly keeping track of their own health at all times

[@sanquinteros]

dude, whats the model?

[@alexclifford2485]

This is outstanding analysis. Thank you. Am impressed with my WD and HGST drives so far

[@sometechguy]

Thank you, appreciate the feedback and the likes on this comment. Glad people found this valuable. 🥳

[@kevinlsims7330]

I Have A WD 500 GB Drive That I Use As A Temp Drive To Record TV To! I Bought It At Goodwill Over 10 Years Ago For $1! This Thing Has lasted Longer Than Any Other Drive I Have Owned And Stays Powered Up 24x7 The Entire Time I Have Owned It!! I Expect To Hear The Click Of Death Any day But it just Keeps on Spinning!

[@yeeaahBUDDY]

@@kevinlsims7330 curious how long it took you to write that comment with all the unnecessary capitalization

[@LatitudeSky]

WD used be my worst nightmare. I had a lot of their 80GB drives and every single one failed. The warranty replacements failed. It was a disaster. But flash forward 15 years or whatever it is and I ended up running WD NVME in everything and now the 10TB HGST HE drives in my NAS. Mostly because they absorbed Hitachi, my old go-to. But I've been quite happy with current NVME and spinning drives and SanDisk for memory cards. Samsung, Crucial, and Silicon Power have all let me down. Would never use those again. Hynix has been good to me.

[@kevinlsims7330]

@@LatitudeSky I had Several 3TB Toshiba Drives That FAILED! The WD 4TB Drives Were Way better! I Am Now Running Refurbished 12TB Seagate Exos 16 Enterprise Drives! I Have 3X12TB and 3X14TB For My OS I Use Sabrient Rocket 1TB And WD Black 1TB NVME Drives! On My retro Dell XPS 410 I Run Kingston SSD NOW 128GB's It's Over 10 Years Old Now! The XPS Is From 2007!

[@csedu3467]

Thank you sir! This is what i was searching. It would be great to find more video like this.

[@TheZettaze]

Thanks for making this data much more understandable, where I've seen other channels make a mess of this, subbed!

[@sometechguy]

And thank you for the comment and feedback. Glad I did a reasonable job and thanks for the sub!

[@anja2440]

I watched 3 of your movies. I like the analysis, and the visual stats about the performance of the drives. You put in a lot of effort visualizing the data, which is most appreciated. It comes in handy to change the WD drives in my synolgy NAS, thanks.

[@sometechguy]

Thank you for the positive feedback. When it comes to failure data, especially comparing many devices I think its really hard to get a good picture of this without visualizations, as there are so many variables to include. It did take some time to put together, and structuring the data also takes time, so I really appreciate the comment that it was useful.

[@davidbon4707]

Thanks for all your work in doing this, much appreciated.

[@sometechguy]

Glad you enjoy it! Thank you

[@CD-vb9fi]

I can't tell you how much information like this is appreciated! It cuts through all the mud. People can make genuinely informed decisions on storage!

[@sometechguy]

Thank you, appreciate the comment and glad it was valuable. 😃

[@frenchfryinyourmcdonaldsba8688]

0:28 ah yes. the notebook and pen taking notes while looking at the back of server racks lovely ain't it

[@sometechguy]

When I am the server room, I always carry a notebook and pen. 😑 But actually, back in the day when patching cables, I really did. But not sure it looked quite like the b-roll.

[@XantheFIN]

Be fair.. notebook and pen is by far easiest anbd quickest way.

[@meateaw]

@@XantheFIN I take photos personally, much quicker and has no transcription errors. (though sometimes they are fuzzy)

[@yw1971]

Somebody please make an app

[@bennysh]

thanks for taking the time making other's choice much easier

[@Collectible_Andy]

Building a nas and found your channel through this video. Enjoyed it enough to subscribe. Thanks for the info!

[@sometechguy]

Thank you and welcome. Small but growing quickly! Appreciate the support.

[@TheKevalar]

an amazing analysis, absolutely shows the difference in manufacturers . allows me to make an informed choice. ❤

[@bioxbiox]

Excellent video and analysis! I am just thinking about building a home NAS with 10+ TB RAID 1 configuration so the video is straight on my needs. Thank you!

[@pl4mbo]

Thank you for this analisys. Great video, simple explanations, looking forward to your content!

[@Jabe_VeX]

very handy video as i go into making my own homelab, thank you so much

[@sometechguy]

You are welcome, glad you liked it.

[@skibik3r]

Fantastic data presentation, the truth is always in the numbers!! I've got about 13k hours on my 16tb X16 drives in my nas, I'm happy to see the data shows they are quality drives!

[@matttownsend7119]

Thanks, excellent analysis, and all of the limitations of the data with their impact on your conclusions have been well explained. I'm just new to your channel but will look out for it in future. I would be interested in an overview of the factors that make Seagate successful despite the dramatically worse reliability shown here - I would have thought that equipment selection for enterprise data centres would be much more sensitive to reliability.

[@johngoodspeed3585]

Excellent video, well researched, much appreciated!

[@sometechguy]

Thank you 🙏

[@stevens1041]

This is incredibly useful. Thanks mate.

[@sometechguy]

No problem 👍

[@bzdtemp]

Love the video. Thank you. And since finding real drive reviews, as those that are to be found tend to be more like, look at one drive in a family and then pretend it speaks for all of them. This then makes your work even more helpful.

[@sometechguy]

Thank you, appreciate this. You can review drives for performance and noise, but can’t really do this for reliability until the drives have got to, or close to their warranty period. And even then, doing this for one or a few drives isn’t very useful. So I think this is the lost useful way of seeing how there drives really perform in a real world scenario, even though BackBlaze may be a harsher environment than some but that is probably a good thing to make sure the drives are subjected to some real work. Thanks for commenting!

[@walterpark8824]

Thorough, and very useful to me. Thank you.

[@sometechguy]

👍

[@cloudmover]

What a wonderful and informative video. Just the facts. Thank you for making some NAS HD upgrades an easier purchase.

[@sometechguy]

My pleasure, and thank you for the feedback.

[@FlorinArjocu]

Impressive work, it is very interesting even for one not working in this field anymore.

[@sometechguy]

Thank you, appreciate you taking the time to comment. 🙏

[@MorningNapalm]

Through the years, my experience has been that Seagates are hit and miss, and WD drives have generally been very reliable. I did have a single WD which failed early, but the rest have been solid. Your videos reinforces these impressions with hard data. Thank you.

[@EnVideoZone]

Great analysis - liked and subscribed

[@sometechguy]

Thank you 👍

[@mateuszbartosik1507]

This just just outstanding!

[@mouldypretzel]

I just watched 5 minutes of this and I have no idea what he is talking about

[@Luredreier]

This is about how likely a number of different hard drives are to fail after x amount of time. It's useful when picking a hard drive to buy for reliability.

[@njdxnjdx]

Totally agree. He should remove this video and I motion for KZbin to ban his posts. Ridiculous

[@jasont80]

This is one of the greatest datasets in all of technology, as it allows our purchase decisions to pressure vendors to build more dependable drives!!!

[@sometechguy]

Yes! I believe this is why Backblaze publish it and why others should to. I am sure it’s leverage over the hard disk suppliers to do the right thing by them also. Backblaze also do a great job at keeping the dataset relatively clean. It didn’t require too much clean up to ensure consistency.

[@jasont80]

@@sometechguy I wonder if we could write an app to monitor drives in Win/Linux machines around the world? That could really grow this dataset. You'd just have to trust the use to determine is a drive failed vs removed. Could work. Would be fairly easy to write.

[@ShinigamiDa]

This is great, thank you!

[@Slugg-O]

That's a lot of info and I know it was a lot of work. Thank you! The Seagate results came as no surprise. They are a huge drive manufacture and a lot of companies and individuals are happy with them, and I'm sure they have made many great drives. Unfortunately, they never made one for me which is why I avoid them like the crackhead at the gas station.

[@Atticman1369]

Awesome video! My 2 HGST 4TB 7200 Red label NAS drives that I've had since late 2014 and they are still going. I'm actually quite surprised I haven't had a drive drop out of my RAID-1. Although the drives are powered down when not being used for 30 minutes which is good, but they do have a lot of spin up counts. Looking for the ideal double digit terabyte drive for a raid 5 configuration with the hot spare.

[@bilujcm]

Thanks, good info!!

[@OvertimeX86]

Thank you for the great video

[@QO0OD]

Thanks for great analysis

[@sometechguy]

You're welcome 👍

[@JeremyLeePotocki]

I got three Seagate Barracuda Drives (Model# ST8000DM004-2CX188) that have been running (mostly) 24/7 for 31500+ hours one of them has finally gave me the caution flag in SMART so I've been needing to get them replaced. I have been deciding on what to get next this video has been a big help. I am planing on getting two 16TB Drives so I can reduce the amount of mechanical drives I have (one in my rig & one as a external backup). All my other drives are going to be solid state.

[@goku445]

Very nicely done video. Thank you.

[@sometechguy]

Thank you, appreciated. 😎

[@hosseinmohammadi4574]

Brilliant analysis. tnx

[@sometechguy]

Thank you 👍

[@mh017509]

Very useful info, thank you

[@paranoidzkitszo]

Great analysis...basically, "if contracted at a company you hate, make sure it's all Seagate!". 10-4-73

[@rollerboogie]

I worked at 2x of these companies. One had me look through backblaze data for an interview question. What id say is that modern HDD are incredibly complex and precise electromechanical devices. Pretty mindblowing they work at all.

[@hopelessnerd6677]

Awesome info! It always intrigues me that in these days of modern times when almost everything is CNC machined to molecular tolerances and the parts should be absolutely identical, that there is this much difference in the failure rates of the different brands/models. Cars are another good example. One person can buy a particular make/model of car and get 500000 trouble-free miles, and another will get a lemon that is in the shop every other week for a new transmission. Doesn't make sense. Hard drives are amazing devices in any case.

[@sigerlion8608]

My 2GB Seagate drive just went kaput after 2 years of minimal use. Even after reformatting and resetting the drive, it only partial transfers/downloads a file and then stops. I now need a replacement and that's what turned me to your comparison videos. Thanks for these detailed videos.

[@sometechguy]

Sorry, sounds unlucky. But thank you for checking out my channel, and I hope the video was useful in helping you compare the manufacturers and products. I have some other videos that look at different aspects of the failure data, but I have not yet covered the disks under 2Tb. I think a lot of people still use these in PCs or for offline backup, so I will take a look that data and see if it provides useful information that's worth digging into. Good luck with the new disk!

[@sigerlion8608]

@@sometechguy Sorry, meant 2 TB, not GB.

[@EJEuth]

@@sometechguyReviewing the 1TB and 2TB 3.5” HDD would be great. For instance: I have almost 10 of them: 5x 2TB (HGST being 2x in a NAS Raid0), 2x NAS 1TB (Maxtor and brand?), 1TB USB (Bufallo). And several (~20 units) of 2.5” from Seagate and WD, ranging from 160GB to 1TB, but mostly in the 320GB-640GB range. ​​⁠Reviewing the 1TB and 2TB 3.5” HDD would be great. For instance: I have almost 10 of them: 5x 2TB (HGST being 2x in a NAS Raid0), 2x NAS 1TB (Maxtor and brand?), 1TB USB (Bufallo). And several (~20 units) of 2.5” from Seagate and WD, ranging from 160GB to 1TB, but mostly in the 320GB-640GB range. As most of these are now “off-line” HDD, for storage or backup, expect for the ones used in LAPTOP’s main HDD (after ~5Y, being replaced by SSD), their actual accumulated hour is typically small (guess to be in the range 1K~5K Hr), but 90% were purchased in 2009~2011. Thank you for sharing such great compilation of information.

[@angrysocialjusticewarrior]

@@sigerlion8608 Its common knowledge that seagate makes unreliable drives (they are hit or miss), and for some reason it is considered taboo to say this even though we all know it. Usually when you say this, you will get a response along the lines of (oh well its not just seagate, all drives have a chance to fail". Trust me, even though any drive has a chance to fail, seagate drives have the highest chance. I feel like seagate has a real life failure rate of 40% after around 3 years but people don't want to admit it.

[@sigerlion8608]

@@angrysocialjusticewarrior Yeah, I barely used the drive. Only uploaded files every few months. Kept it in a cool place. Never bumped or dropped it, and it still failed before reaching 2 years.

[@celtichawks]

Chapters please, saves time if we dont want to watch every detail

[@samsabruskongen]

it's a 15 min video ffs

[@p.p.burnell7294]

Watch the video. Don't be a lazy fuck

[@Voornoek]

ADHD hits hard

[@rt76]

In 7 years using 5 Seagate 4 TB Iron drives in my Synology NAS, one of the five cashed after 3 years and 4 months with bad sectors - shorts outsite the 3 years warrenty. One time after 7 years a drive power off (hang up), but can be reinstalled to the RAID again without an other fail. But in the future I switch to Toshiba, because all Toshiba drives have 5 years warranty.

[@sometechguy]

Thanks for sharing your experiences. The warranty will depend on the drive class, and has varied over time. All the manufacturers appear to offer 5 years warranty currently on current enterprise class disks, but NAS class disks, surveillance and desktop vary. For example the Toshiba S300 Surveillance Drives have a 3 year warranty, and their desktop P300 have 2 year limited warranty currently based on their datasheets. I think generally you will find similar warranties between them on competitive products.

[@suli687]

Best are in order: WD, HGST, TOSHIBA, Seagate

[@Koozwad]

I honestly really like the classic HDD ticking sound my Toshiba X300 12TB makes. Still going strong after buying secondhand a few years ago with regular usage. Also helps to know when the drive is actually being used for diagnostic purposes.

[@sometechguy]

For sure, certain drives have personalities. 😁

[@TheNiteNinja19]

I still have a HGST Deskstar NAS that's been on for 10 years straight, and still cooking along. I wish they were still around.

[@kjm2002]

This was fantastic, thank you very much for your time and presentation. I'm curious if the manufacture site code is not really relevant anymore? If I remember correctly, at least in terms of seagate's, if they had an active recall published, it was generally based on a specified site code in addition to a certain min-max index of serial. As I stare at a few drives here, I'm failing to see an obvious site code on WD, HGST, or Hitachi drives which has piqued my interest if perhaps HGST, WD, Hitachi are all manufactured from a single site indifferent to that of seagate potentially having multiple manufacturing locations...?

[@sometechguy]

I don't believe any of the manufacturers embed this data in the serial number or model number, at least not in a way thats easily accessible. I would love to have that data to include in the comparison, as people comment that the country of origin is related to the quality and reliability. A lot of manufacturer happens in Thailand still, but I think there was some diversification following the issues with the devastating floods that hit that country and seriously impact HDD availability for some time.

[@reecenaidu6020]

two of my drives over the last decade failed at the sata-to-usb adapter. The 1st one was a Vebatim, and the usb end's contacts broke. Taped this one up, copied everything off, and retired the drive. The 2nd one's adapter just stopped working one day. Fortunately I had bought 2 of that Seagate drive while it was on sale, so could test and find the problem. Gonna have to get a sata to usb cable to use the other drive regularly again

[@EinSwitzer]

Looks like voltage retention on the platforms so you will need radial transmission assistance to keep them alogned

[@ianemery2925]

I was really iffy about buying a 14TB WD drive a few years ago; but from your charts, it would seem I picked the best of any brand at the time.

[@bricefleckenstein9666]

If it was an Ultrastar model, that's the old HGST design - WD finally got to merge their HGST purchase from about a DECADE ago into the company a couple years ago. Most reliable drives on the market overall, and WD seems to have been smart enough to have the HGST staff keep designing that line and NOT mess up their production.

[@Jannickjay]

@@bricefleckenstein9666 i would love to buy hgst again, but Cant find it… so hgst is now WD Gold or how?

[@NickDoddTV]

Your video reinforced my own 16+ years in web hosting. While I had strong support for Seagate early on in my business, they've definitely failed a lot more than any other drive. HGST has been my new favorite brand for a while and hasn't let me down just yet.

[@sometechguy]

Thanks for leaving the comment. While, due to the way failure distribution works there can be different experiences, its good to hear that a lot of people share the experiences that the data shows, especially those that have experience with higher drive volumes. Appreciate you sharing.

[@bobbrown8661]

Seagate's are also more competitively priced, hence the larger deployment base IMO.

[@StHabibiJohnsonAhmedFranklin]

Top stuff!

[@landoc05]

My oldest WD Caviar Black is nearly 14 years old, daily heavy use in my main desktop (bought in early 2010, it came with a Seagate drive that failed within the week). Second oldest is 12 years old, secondary drive in that same computer now. Both are working like in the first day, no bad sectors. Both 1TB, older one is a FAEX model and slightly faster access time than the less old FZEX model. I only buy WD Caviar Black 1TB since then. I have a couple more (both FZEX) in secondary machines, both much newer, so far working fine. These newer ones match the access time (13ms) of the FAEX one.

[@jarsky]

I recently upgraded my NAS from my old HGST He8 drives to Seagate Exos. I had 1 X16 arrive DOA (Reallocating Sectors etc...from day 1)..however I dont think it was a logistical shipping issue as the others have been perfectly fine shipped in the same packing foam package. The X20's have been fantastic; no issues with any of them.

[@ChipLinck]

Great video. I have WD, Toshiba, and Seagate drives (and Samsung SSDs) across my NASs. Toshiba is a new addition, but WD and Seagate have been in use for several years. I started with WD over 8 years ago, and added Segate about 5 years ago. Just a hand full of drives. So, no concrete data. I had one of 3 WD fail within the warranty period, which is why I switched to Seagate. Not because I didn't trust WD, but I wanted to place a new drive back in the array same day while I waited on the RMA. Seagate was what was available locally. I ended up getting 2 Segate drives and growing the array. The RMA drive was used elsewhere since at that time I only had a 4 bay and a 2 bay, which were then full due to the new drives. My recent addition of Toshiba drives in a new array was because I will never buy Segate again since they ignored the embargo and sold drives to China. It's no different than selling restricted items to the Soviet Union during the cold war, but I digress. I didn't go with WD after watching a SpaceRex video. I'm shifting my drive strategy anyway. I am slowly getting ramped up to replace my HDDs for data with SSDs, while using HDDs only for surveillance. My new 8 bay model makes it easy to use lots of smaller SSDs and still get the storage I need. I'm growing the SSD array to retire the HDD array with the older drives. The drives individually are cheap enough to pick up periodically, but getting a second 8 bay for backup will be a while. Ultimately, I'd like to have a third 8 bay to place offsite and stop using the cloud for my 321 strategy. It's more like a partial 321+1 strategy, as I use immutable storage for my raw files from each photo shoot. I cull the images, and save the RAWs I am keeping to both immutable, and working storage. I also use the immutable storage for critical documents like my digital fingerprint file (never know when I might want to submit a Form 1 to the ATF), taxes, etc. I finally have enough SSD space to maintain a full 321 strategy, but I'm still relying on the old HDDs as part of my backup, which is not ideal.

[@dociekania]

Seagate ... Advice from 25+ years of experience is: avoid Seagate.

[@dociekania]

Mechanical storage now starts from 4TB up. Personaly I use Adata 8200 pro as main disk, Cruical MX500 for linux, and 4 old 1TB hdd Model: Hitachi HDT721010SLA360 x2, Model: WDC WD1002FAEX-00Y9A0 x2, mechanical drives have about 5.5 years of sumed power on working time and no signs of fatigue so far. In work we have some Synology nas + Synology drives (no idea about model of drives inside) and WD8003FFBX WD Red pro for asustor nas. @@arshkarim_

[@sometechguy]

For the smaller capacity drives, I have a video that compares enterprise drives to desktop drives for failures, found here : kzbin.info/www/bejne/opC8opeMfMh2ZtE. But the short answer is that desktop drives are not generally statistically more likely to fail than enterprise, despite the lower warranty. But the warranty of course allows you to get a replacement for a failure. Yes its true that Seagate appears to have a higher failure rate than HGST (and recent WD drives), and Toshiba can be better also, but its also true that it varies by model, and any drive can fail, or last years. The failure rates over all however, are low, typically 0.5% to 1.5% AFR, meaning that there is around 1% chance that a drive will fail in any given year. So I personally wouldn't avoid Seagate and I have a lot of Seagate drives and have had a good experience. But I would do the following things. 1) Make sure that you do not have 1 copy of data you can't lose on any single disk or in any single drive array. If you care, have a backup. For things I really care about, I have more than 1 backup and keep one somewhere different. 2) If a drive shows signs of failure, replace it. 3) Buy base on pricing, as well as its reliability. I am not saying buy known problem drives, but don't pay huge premiums for specific drives, because any can fail and see point 1. 4) Treat spinning drives properly. Secure them when in use, and don't move them. Don't have them hanging around loose in chassis. 5) Ideally use a UPS to protect systems, especially NAS's. This will result in less unscheduled shutdowns. 6) Be cautious about where you get drives from. I personally buy from trusted retailers where I believe the warranty is valid and its a new drive, the drive hasn't been shucked, removed from a system or 'refurbished'. SSDs are not better than Spinning disks. They are better for certain things. Cheap, long term storage is a good use case for spinning disks and its a more stable medium also for long term data retention. Good luck!

[@GrannyDryden]

Same. Been in the industry 20+ years. I've built a lot of PC's over that time and one thing i can tell you is that when Seagates fail, they fail HARD. What i mean by that, is that, one day all is well and then suddenly the drives start to go downhill very quickly. Seek error climb drastically and if you don't notice it in time or haven't some active SMART Monitoring keeping an eye on things, you'll have a dead drive in no time at all. In Data Centers, this isn't a huge deal, as they have hard drive redundancy as part of their model, but for home users, not as much. I stopped buying Seagate's a long time ago (i can remember when they bought out Maxtor and their plants, so they could increase their warranties from 2 years to 5). I stick with Western Digital personally, for no other reason, that when those drives start to fail, it's a noticeable, gradual downward curve, not a cliff edge.

[@GrannyDryden]

depends on the budget. If you want cheap and cheerful, then go for a Western Digital Blue drive. If you need a bit more speed (like storing your steam library on), then a Western digital Black drive should work nicely. @@arshkarim_

[@club4ghz]

I have 5 WD drives working from 5 to 16 years, none of them died expect one had errors because i dropped it on the floor but i make partitions around the broken area and it's still working.

[@ewenchan1239]

Three things: 1) Seagate vs. WD HGST/HGST WDHGST/HGST is at worst, 1/3rd the AFR of Seagate drives, and at best, almost 1/10th the Seagate AFR. That's HUGE! 2) WDHGST/HGST This is, I think, why WD bought the HGST division from IBM/Hitachi because the drives AREN'T the fastest in terms of read/write/I/O/s performance, but they're absolutely rock solid drives. And yes, whilst you pay more for the initial capex for the drives itself, it also pays dividends with having a lower overall failure rate vs. Seagate. 3) This data shows why I avoid Seagate drives like the plague. 2024 and not all of their drives have some kind of ramp load/unload mechanism for the drive read/write heads, which STILL leads to their R/W heads crashing into the disk platters in some failure cases. This tech is almost 25 years old by now. I remember when IBM first introduced it in 2000 (which then became HGST).

[@LatitudeSky]

Agrees. Used to buy Hitachi drives and now buy the HGST WD drives and they've been extremely reliable. Knock on wood. WD NVME Black has also been excellent. Really hope they maintain quality after the current split.

[@Mtaalas]

Did back-blaze include WHERE those driver were used? I mean a drive that's on a very high utilization server has completely different failure rates than server that's for daily backups which is different from one that has weekly dumps etc... That would be interesting data to take into account... Basically, IO/day data? :)

[@dorkultra]

i've always had good luck with HGST and Seagate Enterprise capacity/Exos drives, even if they are refurbs i'll continue to buy. just make sure you plan for enough redundancy and backups

[@sometechguy]

Yes, no matter which brand you buy or how good they are, they can fail and will fail. And if you are planning for those failures, it can then make sense not to pay a high premium for the most proven reliability also. A lot of people share their bad experiences, and more than most it seems about Seagate. And though the data shows they are probably more likely to fail, I don't think the numbers are too outrageous, and it will come down to statistics. Personally, I have had many years of good experience with Seagate also.

[@GetOffMyyLawn]

Does backblaze track idle time vs usage? As a home server user, my drives are spinning 100% of the time, but active use is probably much less than 12 hours a day.

[@sometechguy]

I don't think SMART has any stats that tell you that, and the data is basically daily SMART extracts along with data on when drives died. So you can see when a drive first appeared, when it disappeared and if it disappeared due to a failure or not. But drives themselves don't record usage, other than number of power down events etc, for which there are very few, probably maintenance. That said, these will be in chassis in large arrays, so I would imagine they are being accessed pretty much 24x7.

[@GetOffMyyLawn]

Ya, I guess we can view their info as a torture test weeding out the weakest drives. All good info... thanks!

[@DevilbyMoonlight]

I still have old SCSI and IDE drives from over 30 years ago and the data on them is still good, 4 of them were spinning constantly in a BBS for over 8 years, the only drives I have had fail in the last 20 years are seagate.

[@mph8759]

Absolutely great analysis and thanks for creating this video for it. Is there an analysis/conclusion (=recommendation) for 4/6/8TB drives, eg. Seagate Iron Wolf? Edit: I’m a subscriber now

[@sometechguy]

Thanks and welcome! I did make the following video that covers the 4-8 TB disks in the data set : kzbin.info/www/bejne/opC8opeMfMh2ZtE it has all the drives and also compares enterprise failure rates to desktop as this dataset has a pretty good mix of both. Hopefully it has useful info you for in it. 🙂

[@mph8759]

@@sometechguy it’s super useful. Thanks so much

[@VTGGT]

The Toshiba's Mg series so for have been very good. Having 2*16TB of the them working fantastic.

[@sometechguy]

Good to hear. They have a smaller market share and there are less vocal defenders and critics of them. But generally, the feedback seems to be positive. The fact that you don't hear a lot of people complaining about them has to be a good sign. 👍

[@VTGGT]

@@sometechguy I also have the toshiba;s HWF. 2x8TB that show up on backblaze's charts. So far no fails. But they are extremely LOUDER than the 16tb's I have now. I also have to mention that My WD 10EZEX the legendary 1TB is 10years old and still working like hell

[@Bluelagoonstudios]

Our 8 poweredge Dell are using the Seagate 10 TB EXOS drives x 6 for each server, and the 6 poweredge are there for backups these have 12Tb exos drives. But after three years, every drive is replaced with new ones, except the backup servers those run 4 years. Our servers are used for content creation, music, video etc. So very big file transfers. For my personal NAS I use the Ironwolfs Pro disks. But that server doesn't do much.

[@mrdali67]

Pretty amazing data crunshing. So Seagate is a good choice when price on each drive do matter, but you also as a former IT professional want some high quality drives. seeing the Exos drives are actually very reasonable priced for an Interprice line of disks but how much does all the different models mean to the individual type used ? I assume that large scale server farms will be using SAS versions only. But I still asume that the SAS/SATA versions is only the interface and the base design of a particulary model and size variant still behave somewhat the same. I had one bad Exos ST18000NM000J which was DOA and am now trying to figure out which brand / model to choose for a 6 disk DIY home NAS project but its a jungle with all the different versions many of the brands have for different use cases. Does it make a lot of difference which specific type is used for a home project ? where you propably have a very mixed load type of both large size videos, pictures and also a mix of very small to medium sized files for games, documents and other stuff. Is it worth coughing up the extra money for a WD if you are already going to choose an Enterprice grade disk ? I have been around for some years and had my shares of both good and bad examples of most of the brands and know that crashes is very random at just about every disk you purchase. My favorites is still Seagate and WD. Beside that single Exos neither Seagate or WD ever given me headaches but IBM/HGST and Toshiba is the only brands where have had disks completely die on me without any chance of normal software data recovery.

[@jackcameback]

Brilliant Stats! I guess that at some point there must also be a tradeoff between reliability and cost? Or is the cost so marginal between brands that it is not considered by data centers?

[@sometechguy]

It varies over time and by disk model. But Seagate often have a price advantage over WD, with Toshiba varying. And yes, I am sure the price vs reliability is going to be part of the decision, especially for customers who carry spares.

[@christiankrueger8048]

Thank you!

[@sometechguy]

Pleasure, thanks for watching!

[@captainhappy]

This data itself can be super boring thing for 99% of us, but you managed to make some sense out of it, thanks for the informative video, there really is some real value in it! I will be certainly looking this video again, at the time when I consider my options upgrading harddrives in my NAS.

[@sometechguy]

Personally I love mining data to find hidden information, but it isn't for everyone. 😁 Thanks for checking it out!

[@DJdoppIer]

The DOA / dead in weeks rate for some of those Toshiba drives are insane! Sure you can get them replaced under warranty, but that's an annoying hassle for people that need to deploy them ASAP. I gotta wonder what kind of (if any) QC goes on at Toshiba.

[@Stoney_Eagle]

Seeing my experience backed up by numbers is fantastic! I have had so many hard drives fail from many different brands, and have lost a lot of data because of it. I have yet to have a Western Digital to give the ghost on me without complaining about old age first for months of slow speeds, screaming "Please retire me" 😂 Many drives die by looking at it wrong but have rough-handled WD's, and they are still alive today. This is how you earn my brand loyalty! PS. Don't worry I have grown up now and backup properly on raided systems.

[@paulmichaelfreedman8334]

Yes, I've been a long time WD user, because all this time, to me, they've proven to be most reliable, overall. In my box of antiques I have one of the first SATA drives, a WD1600, from 2003. I can plug it in and use it if I like, even now. Still works perfectly thanks to the FD bearing. In fact, all the WD drives still work and I have a lot of them, accumulated over the years. HGST also performs very well, and Seagate...well...I've had plenty of those conk out on me. Quantum drives never failed me, until Seagate raped them. Never had IBM drives, they tended to have bad batches. When it was bought by Hitachi, quality went up dramatically (This is now HGST). Moderately satisfied with Toshiba 3.5 inch drives, their 2.5 inch laptop drives are exceedingly good.

[@m1stertim]

RAID is not a backup

[@Stoney_Eagle]

@@m1stertim I think you missed the s there at the end. 😉

[@walkman1269]

I've always FELT HGST and WD Enterprise were the bomb. I see that my gut feeling was right. I had a 4TB die once but it was consumer grade.

[@sometechguy]

Unfortunately, any disk can die and for different reasons, so backups are always important. I get a variety of comments on videos about various vendors and people are most likely to hate on products that have failed on them understandably, but WD seem to have broad respect for their HDDs at least, and the data seems to support that. So yes, your gut isn't failing you! Thanks for commenting and sharing. 😎

[@wendohgermaine6448]

Hey great video.... Would you suggest one to get Western Digital over the rest and especially enterprise disks as I think enterprise disks laster longer. I want to build a fairly large NAS, and seeing I've had a few WD black enterprise disks for the past 8 years they haven't failed on me once.

[@sometechguy]

Hi and thanks for watching and commenting also. I put out a video a couple of weeks ago which covers exactly this I think, found here: kzbin.info/www/bejne/n5Sxk5iujrtmg5Y. I hope this gives all the specific data and context to help. The TL:DR is that WD disks appear to currently be the most reliable enterprise disks, and although the available data for these disks only goes back 2.5-3.5 years or so, it appears to closely follow the past behavior or the Ultrastar range when it was branded under HGST. The small caveat is just around price point and availability, but I just bought some 20Tb Ultrastar DC560s myself as they were actually cheaper at the time than Exos 20Tb disks. Toshiba Enterprise Capacity 20Tb disks were cheaper still, but the current data seems to look great for those Ultrastars, so I went with them. But of course, price varies day to day and by location, so your mileage there may vary. Good luck with the NAS build.

[@bricefleckenstein9666]

@@sometechguy HGST / WD "Ultrastar" have been my go-to models for about a decade. Followed by Toshiba enterprise models.

[@jfox8888]

So.... after going through the video, thank you for the free content and the effort put in. the conclusion im getting is : WD first, HGST [ unlikely avaliable], followed by toshiba ? my comment is that : theres no take away point from the entire video, if im a layman, going for what is the recommended. Yes theres not much further data to infer to, but there can always be disclaimers given that things go wrong, and will have higher chances as the size increases

[@sometechguy]

I prefer to stay away from telling people what they should do, and instead provide data to make their own informed choice from. The overall reliability of the drive is not the only factor, price availability and reputation of the vendor is also something that varies over time and location. Some people are not interested in the detail and just want to be told which to buy, others will want the data and make their own choices. So its a balance. But I take the feedback that the conclusions could be more concise. Creating content is about learning and improving with each video, so this feedback is constructive and useful. Thank you for that!

[@woobilicious.]

My friends "Big foot" either died, or got retired a few years ago, if you're not familiar with those, it's a 4GB, 5.25 Inch Hard drive using IDE, from the 90s. Sometimes there's a golden sample that outlasts Middle eastern democracy.

[@andynonimuss6298]

I have several 15 year-old Western Digital drives that still work just fine.

[@sometechguy]

This is good to hear. The data shows they are excellent drives, and I did some other analysis also comparing the manufacturers more broadly and HGST and the WD descendants come out glowing.

[@kingneutron1]

Thanks!

[@sometechguy]

Thank you for the comment, and especially appreciate the Superthanks. 🙌

[@sovahc]

Thanks 😁

[@leaf16nut]

I’ve always been a WD guy, haven’t had one die in the last 15 years of PC gaming, but 12TB Seagate drives were on sale and too good to pass up, only using it as for media storage with not a lot of reads or writes so hopefully it lasts 😫

[@blahorgaslisk7763]

This was a long time ago. A representative for WD talked a bit about the test and quality of different drives. He told me they had tested the cheaper drives to the same standard as the professional drives. The reason they put so much effort in the cheaper drives was that at the volumes they shipped a single failing cheap drive would eat up the income for something like ten other drives. Meanwhile a professional drive that failed and had to be replaced would only eat up the income for something like five or six other drives of the same kind. So the reliability of the cheap drives were actually more important financially. It kind of broke my idea about how the big drive manufacturers worked. We kept selling enterprise drives to our customers even after that. It wasn't worth shaking the tree that hard.

[@jjptech]

13:46 did you made that? Or is skynet who did it?

[@jaapkamstra9343]

Great review! Great to see how my 14TB wd is stacking up.

[@mazz85-]

Suggestion for longest storage life for videos ? No gamin or anytin else installed or used.

[@babthooka]

Good video.

[@sometechguy]

Thank you 👍

[@freevideos051]

I had the best luck with Western Digital drives, the first drive I bought was in around 1997.

[@H0mework]

Are the early high failure rates common for helium drives? Maybe it's damaged in shipping.

[@sometechguy]

If anything drives have been improving in reliability. The very first he8 drives took a small hit, but drives vary so can’t be sure it’s related to the helium fill process. But generally I don’t see a reason to believe that helium drives are less reliable. But transit damage is a hidden enemy in drive reliability.

[@pikapika3]

Greetings, im struggling between choosing one of this drivers, if you could help. its for PC but will be using for media server, so work constantly EXOS™ X16 10TB ST10000NM001G and 10TB Toshiba S300 / MG06ACA10TE

[@sometechguy]

The Enterprise options (Exos X16 and Toshiba Enterprise Capacity MG06) will come with 5 year warranties, which is a positive but will be a bit noisier if its for a PC. So depends where the PC will be running if the noise is a consideration. The S300 and the S300 Pro will come with a 3 year warranty I believe, and will be a bit quieter, the S300 doesn't come in a 10Tb AFAIK, so you would need the pro version to get that capacity, and the Pro is a 7200RPM drive like the X16 and MG06, where the S300 non-pro is a 5400 RPM drive. You could look at the N300 also, which is a NAS drive. I don't think there is a lot between them. The X16 seemed like one of Seagate's better models, and the Tosh drives also are good. So it probably comes down to price, warranty period and maybe if the noisier drive will be a problem for you if its running 24x7. But the usual caveat is that any drive can fail, so no matter what you choose, that can happen and plan accordingly. 😉 Hope this helps you make your choice!

[@pikapika3]

tnx, im just skeptical about Seagate drivers over the 10 years 3x6TB ones died on me.

[@nocturnal101ravenous6]

@@sometechguy The Toshiba drives are loud, I am using the 2x X300 Pro I don't know how they compare to the N series, but they are pretty loud, honestly the X18 EXOS drives(Both are 16TB) while they are a little loud they are not actually that bad for a PC, I am using a Lian Li Dynamic Mini so results may vary depending on case, If you are using something like a Be Quiet Dark Pro case you are probably not going to hear anything.

[@court2379]

These figures are for continuous use, correct? If the use is intermittent, say six hours a day, does it it extrapolate to 4x the life? Or does the way its used increase or decrease the MTBF?

[@sometechguy]

The use case is 24x7, but the test data is actually calculated against power_on_hours anyway and not elapsed time. So its based on continuous use, and because the power on hours lines up with the drive ages closely, its clear they are powered on for approx 24hours a day on average. I am sure there are brief maintenance windows. But in terms of the stated MTBF on the drives, its based on assumed workload for the drive. For enterprise drives in this video, they are all 24x7, but MTBF on desktop drives is often based on numbers around 8 hours a day.

[@blahorgaslisk7763]

Back many years I remember Seagate having a series that was very low quality. The failure rate was very close to 100% over a year. Yet we had a customer who never let us replace the drive he had. Instead we opened it up and nudged the read write head and closed the drive back up again. It would continue working for 3 to 6 months before failing and that customer would be back again. It was incredible that the drive kept working. From all I knew about HDD's it should have crashed the head, but instead it just worked. Another drive I had, a Maxtor I think it was, was also strange. A metallic sticker had been torn off opening a hole in the cover of the drive. The disks pushed air out of that hole so I could feel it on my fingers. It also made the drive sound strange. That drive I swapped out for a customer who wanted a larger drive. I kept using it for test installations of windows on machines and it just kept working even though I expected it to die at any time. But for a few years I used it over and over and it never gave up. Finally we sold a crapload of IBM "Deathstar" drives. I'm talking about several hundred of the worst of the worst model. And yet we had less than ten that came back before they were out of warranty. It was strange that we had that much luck with that drive that was documented to have extremely high failure rates. We had them in our own server, and in six years none of those died.

[@paulmichaelfreedman8334]

ST3420 or 2340 perhaps? I remember those drives failing in bunches back in 1995. Some IBM drive cases would warp above 45C and crash the heads.

[@sometechguy]

People complain about disk failures, and its understandable because they can have a really serious emotional impact to people when they contain important data like photos, especially if not backed up. But considering the incredible precision engineering and moving parts, they still often last 5-10 years. I just did a video on the engineering on the new Seagate Mozaic drives, and its amazing the things work like they do.

[@pooiyx]

So if I want to buy a 10TB enterprise HDD, WD is the best choice?

[@sometechguy]

There is a short and long answer. Short Answer: WD appear to be producing more reliable drives over the last few years. And across the models I analyzed, that seems a solid trend. I did some other videos on this as well, and the data seems to robustly support that, even if you slice it different ways. Long Answer: Drive reliability varies by model, and its all statistical based on the available sample. So any drive can fail, but the data indicates the likelihood for each model. Also, the 10Tb models analyzed in this data set, may not be the ones you get today. Actually, you can pick up 10Tb from different model lines and reliability could vary, for example between an Exos X10 10Tb and X18 10Tb. But taken in aggregate across a very large sample, WD appear to provide the most reliable drives. Its also worth noting that 'most reliable' doesn't mean the best drive to buy. Price is going to be a consideration, as well as how long you plan to keep the drive, and your trust in the brand as a whole. Support and warranty is also important, though I don't think any of these manufacturers have problems with their warranty, as long as you buy from an approved reseller.

[@lixnix2018]

I have 20TB Toshiba drives in my storage server no problem so far. But I know from working in a data center that array rebuilds can be dangerous for drives.

[@federicocatelli8785]

What would you get in the 10-12 Tb ?

[@EinSwitzer]

Structure and even thickness changes

[@velsvelsy2656]

Would you consider sharing the analysis result?

[@sometechguy]

The analysis is done on the original Backblaze data, which is really large (67Gb currently compressed). Each of the data points in here is derived from SQL queries on that data, so the working data isn't that easy to share. But Backblaze do publicly release the data source if you want to dig into it. Available here : www.backblaze.com/cloud-storage/resources/hard-drive-test-data

[@EpicLPer]

Most of the drives I've ever had fail in my life, be it my own drives or from people around me, were mainly Seagate, and that for like 15 years minimum now... why is that company still around and what are they doing...

[@Guanaalex]

I built my first PC in 94’. The Ultrastar was a SCSI at that time and the best drive on the market. Not surprising, your test confirms it for todays drives. Great work on those numbers.

[@sometechguy]

Thank you! SCSI was a pain with those dedicated SCSI cards and those 50pin connectors, but there was also something satisfying about it. 😊

[@oncrei]

Weibull statistical analysis is what you need to analyze this data

[@kusonsin]

Interesting if all these drives are randomly placed in the data center, or whether some brands are used in DB / some are used for cold file storage etc. ---- random is the assumption, but if the 'real usage' time could also be displayed in. But perhaps that is confidential

[@sometechguy]

Backblaze just started to add the storage pod data to the source, though historic failed drives likely don't have it. I need to go into the new dataset and see what changed. This will reveal if the drive is part of their normal customer offering, which is just pooled storage for customer archive. And for this use case, the type of usage should be pretty uniform. As I believe they use different disks for OS instances, and would think for DBs that they use fast storage but might depend on the case. But I think in any case, its a safe assumption that the vast majority of the drives in the data are used for their cloud backup service.

[@LokiDaFerret]

As an IT guy this is really handy information! So you get a thumbs up for that. 👍 However your presentation of this information is a bit lacking. It would be far easier to read the graph if you focused on one vendor at a time getting rid of all the noise from other vendors.

[@sometechguy]

Appreciate the feedback. There is always room for improvement and the balance is comparing the models from each, to how the manufacturers compare to each other. It’s been a while since I did this and there is more data becoming available, including 22Tb drives, so if I do an update, I will see what I can do to improve on the visualisations. Thanks for commenting

[@LokiDaFerret]

@@sometechguy What gravitated me to your videos is the fact that I need to buy some new drives for my NAS. So my focus is not so much on any particular model but more determining which vendor is the most reliable generally speaking. Let's face it any drive you buy is a crapshoot as to its longevity. You need to find a happy balance between price and risk.

[@sometechguy]

@@LokiDaFerret you may have seen it, but I did a video comparing the vendors where the data is cut into models lines vs individual drive models. For example, Exos X14. X18 or Ultrastar HC520 or Tosh MG08, and for the reason you say. Thanks again.