วีดีโอ

The first thing to say, is that any drive can fail and it can be drive quality, which is probably the difference between a large dataset like the one I used, or can simply be bad logistics such as packing and handling. But overall, it seems statistically, WD will be most reliable. That said, none of the manufacturers are so bad as to 'avoid', but I consider this depending on the use case along with the price. I have quite a few Exos and Ironwolf Pro's with no issues, and I have quite a few Ultrastars also. I had WD Reds fail, but I just had quite a few and they were over 7 years old. So I wouldn't call that a major issue either. There are some models to avoid, and it seems generally the newer models seem better. Those X16's seem pretty good. I didn't have as much data on the later models. But you are looking for cold storage, you can probably focus on cost per Tb and not get too worried. I have all my data on RAID using Pro or Enterprise grade disks, and my most important data I have two copies on two different NAS; but my cold storage backups are on lower end 'cheap' drives, often external disks that contain drives like barracuda's. Those don't have the best rep but I had no issues, even on very old disks, considering how little time they spent powered on, and are cold backups they are not at serious risk of failure.

​@@sometechguy Thanks a lot for your exhaustive answer! I totally agree, every disk can fail and that's why I use to have at least a second disk as backup. Anyway since I would use also to work on it and the disk would likely be under load quite a few hours every day I'd like to go for the safest bet. I thought you mentioned on another video that the 12TB version of the Exos had an history of higher failure rate and while the X16 is on average quite more reliable seems also not easy to find this series (may be discontinued?) but rather in the X18 or X20 versions. All in all, having to chose between 12, 14 and 16TB disks do you think there would be one that you would prefer for reliability (perhaps being better engineered)? Thanks again for your time!

Exos X series anyway should be entirely CMR, isn't it?

@@twowheelslife5611 yes, the Exos range is CMR, with a couple of very specific disks that are host managed SMR, but you shouldn’t be able to get those by mistake. The data didn’t include much on the X18 and none on the X20 but I personally bought some X18 18Tb disks. Though it isn’t going to be true that newer disks are always better, there does appear to be a trend of improvement. So I wouldn’t be put off those disks until I see data. Bit more recently I have been buying the WD Ultrastar as they seem the best for reliability.

@@sometechguy Thanks again for your reply. If so shouldn't be any difference in terms of reliability closing one size or another, thus going with Seagate any 12 14 or 16 TB version will probably be the same. Could be due to my bad experience but I still have some bias toward WD, anyway I'll still consider this Ultrastar in my choice before to invest in set of them. 👌🏻 Thanks!

For home users and SMB, this would be my opinion. For larger enterprise use cases, the cost savings are probably beneficial to run the Exos and you just carry spares, replace any failures and warranty the drives. Which you would have to do anyway, no matter which manufacturer you choose. But it probably just comes down to cost and the warranty process, which is probably via your server provider anyway.

Thank you, and my pleasure to help.

The black may be a little quieter, as the firmware is more tuned towards that and less for performance and 24x7 running. But from what I have seen on the drives, it’s likely the same platform/hardware with some firmware changes. NAS and enterprise will sleep less often, though you can set lower save settings, and maximising performance often means more aggressive seeks, which makes the drives a little louder. But personally, I would compare pricing and if the Ultrastars are the best price per Tb, I would go with that. The prices may vary, but often the Red and Blacks are more expensive than the Enterpise drives.

😭

Conner and Maxtor also acquired by Seagate eventually. I remember both of these back in the day when there wasn’t just three HDD manufacturers.

I tend to think the blacks are over priced because WD positions it as an enthusiast or gamer drive, and they are usually willing to pay more. I would personally look at things like the Ultrastar, gold or red pro , or even purple pro and see how those compare for price. I did a video also on the WD range and how they compare for price to TB, may be worth checking that.

@@sometechguy thanks a lot, let me watch your video. Yes? I see WD are are over priced. in this case even the WD Black 6 tb and the Seagate 8tb Barracuda are al 130 dollars price. Is the same price. Even though, I see many people say the WD Black ir really noisy. What en yet think. However, let me watch your video. Thanks a lot for your answer.

@@sometechguy Tnasks a lot, I am watching your video.

Thank you, appreciate the feedback. 🙏

🙄

@@sometechguy 😉

The failure rates for both are not super high, so if you are a regular disk buyer or even someone that buys quite a few drives over the years for home use then it’s quite possible you get no failures for the useful life of the drive. But you could always suffer an early failure. So depending on which you buy and your luck, you get either a good or bad impression of that manufacturer. I see a lot of comments on these videos about people who hate either Seagate or WD, or Toshiba because of those bad experiences. Seagate get the worst press overall, they do have higher fail rates but also a much bigger market share, so more chance for someone to have a bad experience. But in the end, I think it’s just chance. Unless you really analyze the data and find out what the statistical probabilities are. And this is what I did here. If you do buy a lot of disks, you get a more realistic picture but you can also factor in cost and warranty process, as Seagate are often (not always) cheaper to buy. So if you save 10% say, but get 10% more failures, which you can return under warranty, it’s probably a good deal over 1000s of drives. Especially if your servers get replaced before that warranty expires.

They are Toshiba drives, rebadged and with some ‘firmware updates’. On principle, I would not buy them and the more Synology push me towards them, the less likely I am to continue to use their NAS. As you might tell, this irritates me. So I am not planning to look at them myself any time soon. :-) I would just encourage no one to buy them. A small uplift on price might be reasonable, but it’s an outrageous inflation in price in my book. 😞

They were discontinued about 9 years ago.

On the face of it, it doesn’t make a lot of sense. But if you look at it more along the lines of ‘if I have 285 drives, each year 1 will fail’, it seems more credible. But one of the problems is that these are predictive estimates, based on failure rates of components, and also they don’t consider the life of the drive. As you point out, it’s very unlikely any drive would last 285 years. Once the warranty period is past, you can certainly expect rates might start to climb. And actually, other than some specific models, most drives here never achieve their stated MTBF at any point in their life.

Thank you 🙏

This is the way

I couldn't find a HC510 for sale easily, but if I compare the 12Tb HC520 and the 16Tb HC550, the price per Tb for the 16Tb drive is maybe 20% better. The data I have seen suggests the newer (larger) drives are also a little more reliable. So personally, I would pick the 16Tb. The premium in the UK right now, is around £30 including tax, so around $38 US. So if you can afford that premium, its better value per byte. Of course, not sure on price and availability where you are. And thank you for watching, even though English is not your first language and I hope you could get some value from the video.

@@sometechguy I was thinking of buying toshiba mg 10tb. Then, thanks to your video, I thought about buying Ultrastar 10TB, but when I learned that the HC510 series was also produced by HGST, Then I decided to save a little more money for the Ultrastar HC550 16TB, which is the original WD. Thank you sir.

WD owned HGST for a long time now, and now all the HGST Ultrastar drives are branded WD. The HGST drives had a great reputation for build quality and reliability, and that still appears to continue now that they are branded WD. Good luck with the drive purchase!

@@sometechguy Oh I thought only ultrastar hc510 ones were hgst. Now I have to think again. Thank you for your time and information andd good luck on your TH-cam journey!

There isn't data for either of those specific models, but based on the holistic data from drives in those ranges, the Ultrastars look to be the most reliable. But, as I always point out, that doesn't make it a definitive choice. The Tosh MGs are often quite a bit cheaper, and they still appear to be good drives. And they all come with the same 5 Year warranty also. But if its for home use, and you don't run large numbers and carry spares, and drive failures are likely to be a real inconvenience, I would pay the extra (if its reasonable) and get the Ultrastars...personally.

Thank you 🙏

Thank you, appreciate it. 🙏

I totally understand. I did some analysis and they do appear to be shipping the best built and reliable drives, especially in the enterprise drive class, which is also often the most efficient. So I have been buying some of these recently, but I tread very carefully, thoroughly check specs and have to hold my nose a bit. I do also have a video I am putting together currently comparing their Blue range, which contain SMR and CMR units and 5400rpm 7200rpm spindle speeds, which are still very unclearly marked, so see how they really compare for performance.

I am sure they will serve you well. I have quite a few 18Tb Ironwolf Pros in an array. But for the reasons in the video, I changed over to Exos on that NAS, as they seem way better value.

The dataset doesn't include any high capacity NAS drives, just Enterprise drives. So there just isn't any data there on those drives to analyze I am afraid. I have another video comparing 4, 6 and 8 Tb drives which contains Enterprise and Desktop drives and this dataset has a fairly small sample of WD Red drives, but the numbers are pretty small, and they did not fare well, being decommissioned pretty early following poor reliability. That video can be found here: th-cam.com/video/l_YqdVGcC0o/w-d-xo.html

So 'Great work' but not use to you at all? 🤣 Sorry if this was the case, were you viewing this on a phone or on a smaller display? I think even if so, its good feedback for future videos to try and make this more accessible. So appreciate the comment and the feedback, and the sub of course also. More content in the works and hopefully you find it all valuable.

@@sometechguy Oh no it was definitely usable with some rewinds and mostly listening to you. But if I wanted to just pause the screen and read the graph for myself, as I personally often do, it was a pain in the butt to read. Doable tho. It was great work overall and I super appreciate the hard data driven approach, and the scientific literacy you appear to bring to the table. I was able to get useful info. I hope my first comment didn't come off as rude and I'm genuinely sorry if it did. You know I meant to tell you lol I actually tried looking at your graph first on a 27in 1440p, next a 55in 4K, and I'm now checking next on my 1080p phone. So looks about the same except now the text is a struggle to read, as well. I could put my phone in 1440p mode and maybe see it a bit better, but battery life... I can see the colors maybe a bit better or the same in the graph itself, but crucially not the legend/key. Those thin little lines next to model numbers mostly all look the same. I wasn't sure it was supposed to be legend/key at first glance. I will admit I am recently 40 and starting to need reading glasses for particularly small stuff, so take my single data point with a grain of salt. I've already watched a couple more useful videos, btw. Great to have another data driven channel to go to :)

Thanks, it’s really useful feedback and as long as feedback is constructive and actionable, I always welcome it. I have some other videos I am currently working on, but will be doing an update on this one before long as there is updated data now available that contains more of the 22Tb disks. So I will certainly to consider this kind of feedback when I work on the visuals for that one, and other videos.

I am sure there is a reference to a cult called SHR that I am missing here. 😌

The best explanation is probably the cost. If the drives are significantly cheaper and the operating model is to carry spares and replace failures which have data protection and redundancy in place, it might still save significant money. Especially if the drives have 5 year warranty and will be replaced if they fail. The decision for smaller deployments could be different though, where drive failures could be more impacting. And though the failure rates, at least in this data are higher, they are still fairly modest with most drives surviving well past their warranty period.

It will depend on data volume, but RAID should never be an alternative to a backup, it’s an extra real time mechanism for data protection. For example, if you backup to an external disk, there will be a delta between your live data and your backup, which could be a day, or many months, depending on how rigorous the backup regime is. For portability, it depends but you should usually find moving an array between similar hardware should be fine. I once had a mainboard failure on a Synology NAS and the unit was replaced, the disks went into the new unit and the entire thing started like nothing has happened, NAS OS, all volumes etc. it may often be recoverable on other hardware if the software is the same also. But recovering SHR on another device than a Synology is more complex than a basic RAID config. And yes, replacing a failed disk and recovering an array is a high stress event for other drives. It’s totally fine for them, but if any of them are also verging on failure, it can kill them and I have seen this happen. RAID 6 or SHR2 on Synology reduces that risk, but before I do a RAID recovery, I check all my backups first. If you don’t have the high capacity requirements, what you are doing is good. Depending on the criticality of data, another backup kept elsewhere is also good. Maybe not for everything, but the things that really cannot be replaced. Thanks for watching and for the comment, great questions!

Drives normally state their read/write in MBPS, so this is a starting point. However, this performance is often based on either SLC cache speed, or if they have DRAM. Once the cache is full, you are down to the read/write performance of the main NVMe storage. If it’s QLC, it may be slower than 1000MBPs. It’s a bigger topic that may need some research, but mostly TLC (Triple Level Cell) NAND will probably be able to sustain the 1000Mbps, QLC (Quad level Cell) may not. QLC offers the highest capacity, but it’s slow and not as durable. So check for both the speed, dig into the sustained speed and check the cache size as the cache getting filled will determine when the sustained write matters. Depending how much you read/write at a time will be the thing to consider. I want to do some content on all this topic also. Have a few things coming before this though.