Seagate Launching Mass Market 20 TB PMR HDDs in The Coming Months

[Uskompuf]

Seagate has recently announced during their latest earnings call that they are preparing to launch mass-market 20 TB PMR hard drives in H2 2021. Seagate already has several business-focused 20 TB HAMR drives which are available in limited quantities to select partners. These new 20 TB PMR drives will feature two-dimensional magnetic recording and are already sampling with customers. Seagate is also working on SMR 20 TB drives for Hyperscale systems with specialized software. Seagate hopes this growing lineup of 20 TB drives will help them address specific customer needs.

We expect to begin shipping 20 terabyte PMR drives in the second half of this calendar year.

View at TechPowerUp Main Site

 

[Blue4130]

Perfect plex drives.

 

[Jack1n]

Just had a 2TB drive from Seagate fail on me without warning. I doubt I will buy another mechanical drive ever again and if I do, it will not be from Seagate.

 

[TumbleGeorge]

Little old but:

Dude, Where's my 100TB Hard Drive? - Architecting IT

Hard drives continue to increase in capacity. So how long do we need to wait until we get the 100TB drive the industry has been promising us?

www.architecting.it

CAGR: "26TB HDD in 2021", where is?

 

[R-T-B]

Just had a 2TB drive from Seagate fail on me without warning. I doubt I will buy another mechanical drive ever again and if I do, it will not be from Seagate.

HDDs fail, from all brands. Don't base your experience on one drive.

I had a Toshiba do the same to me (no SMART warnings just died) but I still buy Toshiba if there is a relevant need.

 

[Jack1n]

HDDs fail, from all brands. Don't base your experience on one drive.

I had a Toshiba do the same to me (no SMART warnings just died) but I still buy Toshiba if there is a relevant need.

It is not just my own personal experience, it well known that Seagate drives have a high failure rate.

 

[R-T-B]

It is not just my own personal experience, it well known that Seagate drives have a high failure rate.

I've never seen stats to validate that very popular forum rumor, but I have seen stats that sort of go against it, ironically.

So same point. I think that rumor started with their early/first perpindicular recording drives, which DID have a high rate of failure, but it has since been more than a decade from when the last of those particular problem batches were discontinued.

 

[TumbleGeorge]

@backblaze.com

 

[JAB Creations]

Just had a 2TB drive from Seagate fail on me without warning. I doubt I will buy another mechanical drive ever again and if I do, it will not be from Seagate.

Were you running RAID 1 and regularly clone to an external drive?

 

[Jack1n]

Were you running RAID 1 and regularly clone to an external drive?

No, but I had the important stuff on my Google drive. Everything else is replaceable but it is a chore.

 

[mashie]

Any guess what the price will be for these? Not that I need to replace my 10x 10TB Seagate IronWolf based array anytime soon.

 

[Mindweaver]

We can't get bigger drives fast enough for me.. and much cheaper.. lol I mean we have games that are north of 100gb. I did pick up 2x 1tb ssd's for 65 bucks during Amazon day or whatever it's called. I want an 8-10tb drive for under 100 bucks.

 

[Octavean]

Well, chia farming or some other form of natural disaster has caused HDD prices to skyrocket.

I bought six 14TB WD drives some time ago and those same drives are now twice the price. I was also sporadically buying some cheap enterprise surplus 4TB Hitachi drives for around ~$50 USD and those have risen to about ~80 USD last time I checked. New 20TB HDD's are great and all but if the pricing scheme has been destroyed by demand and availability its of no use to me.

 

[Purgatory]

Seagate drives make me shudder like the Hitachi deskstar aka death stars.. no thanks no way on earth I would risk 20TB of data on their drives, learned my lesson with both these brands and gave them many chances with different drive models and all started to fail with the click of death, they all fail one after the other. Western Digital Blacks only for me, never had one ever fail even ones running now for many years, even have greens and blues still working to this day.

Hitachi caused me so much pain, I clone my drives and they all were purchased at the same time as I always buy them in pairs as one live drive and one a full backup of the live drive, had 6 drives die on me a week after the first one started one time and couldn't believe it to the point they started to crash my only backups and had to spend days recovering my data. No thanks I would not want to wish that on my worst enemy.

 

[chodaboy19]

Uh... I need 6 of these for my NAS...

 

[Octavean]

I mostly buy WD drive currently (Red and Black) and I've had my fair share of WD drives fail. This is anecdotal but I have purchased from the major HDD manufactures out there and I have had failures from all brands that I've tried.

If someone decides they are never going to buy from a company because of a past failure then they will in all likelihood quickly run out of manufactures to buy from. The best you can do is to strategize against loss of data. Use the 3-2-1 rule of backups and you should be good.

Saying ~20TB is too much data to lose suggest wrong thinking and implementation of storage / backup strategies because no one should be relying on the one drive or even two for that matter,....and RAID is not a backup.

 

[newtekie1]

@backblaze.com

While those statistics seem nice, they are flawed.

First, they don't really apply to the consumer, as they are using drives in environments they aren't meant to be used in. Also, what Backblaze considers a failure isn't what a failure would be in a single drive consumer application. Backblaze labels the drive as failed if the RAID(or whatever redundancy system they are using) kicks the drive out of the array. But when you are using consumer drives in multiple drive array applications, that can just happen even if the drive is perfectly fine. Sometimes it comes down to just the drive not responding to a request fast enough, which wouldn't be a problem in a single drive configuration but is in a multi-drive configuration.

Second, do you find it odd that they only use 2 WD models? In fact, I think for a while they had no WD models. You really have to ask yourself why a company that are running huge numbers of drives eventually just stopped using one brand for a while. And they answer this question, they had some WD model drives with 100% failure rates that they didn't even bother to include in the quarterly stat reports. They've said this on their blog before.

We can't get bigger drives fast enough for me.. and much cheaper.. lol I mean we have games that are north of 100gb. I did pick up 2x 1tb ssd's for 65 bucks during Amazon day or whatever it's called. I want an 8-10tb drive for under 100 bucks.

Yeah , I just put an 8TB for my games and minor other storage in my main PC a few months ago and I'm already down to less than 1TB free.

 

[dragontamer5788]

Backblaze labels the drive as failed if the RAID(or whatever redundancy system they are using)

Backblaze doesn't have a redundancy system at the hardware level. Its their "web-software" that offers redundancy. Your data is stored in 20-different racks, in 20-different computers. At the OS-level, all of their drives are just JBODs (aka: Just a bunch of Drives). No RAID, no striping, no mirroring, no nothing. The web-software "manually" performs a Reed-Solomon 16-data+4-parity calculation.

Which means that if someone were to go into Backblaze's datacenter and tip-over 10% of their racks, you'd suffer absolutely no data loss. In contrast, if you have a computer with RAID and that computer fell-over (smashing the hard drives: Racks are like 1000lbs fully loaded... you really don't want them to fall over) all data would be lost. No amount of RAID protects you from the computer getting smashed: you gotta just store the data on different computers and then recombine them later.

 

[TumbleGeorge]

While those statistics seem nice, they are flawed.

First, they don't really apply to the consumer, as they are using drives in environments they aren't meant to be used in. Also, what Backblaze considers a failure isn't what a failure would be in a single drive consumer application. Backblaze labels the drive as failed if the RAID(or whatever redundancy system they are using) kicks the drive out of the array. But when you are using consumer drives in multiple drive array applications, that can just happen even if the drive is perfectly fine. Sometimes it comes down to just the drive not responding to a request fast enough, which wouldn't be a problem in a single drive configuration but is in a multi-drive configuration.

Second, do you find it odd that they only use 2 WD models? In fact, I think for a while they had no WD models. You really have to ask yourself why a company that are running huge numbers of drives eventually just stopped using one brand for a while. And they answer this question, they had some WD model drives with 100% failure rates that they didn't even bother to include in the quarterly stat reports. They've said this on their blog before.



Yeah , I just put an 8TB for my games and minor other storage in my main PC a few months ago and I'm already down to less than 1TB free.

It's just for orientation. I have not written any comment below the photo because I do not claim anything.

 

[dragontamer5788]

Second, do you find it odd that they only use 2 WD models? In fact, I think for a while they had no WD models.

For a while they had no Toshiba models because Toshiba put the sticker on the wrong side of the hard drive (even though Toshiba had high-reliability rates in the small-scale tests... they didn't want to manually place a sticker / use a label maker).

Don't overthink it. There's lots of mundane reasons to cut out a manufacturer. Sticker on wrong side is one such reason. I see that they have Toshibas now, so maybe their newest storage-pods have stickers in the correct spot.

Alternative explanation: Backblaze popularized the 40-HDD "vertical" orientation for 4U servers. Years ago, they'd build their own servers. Today, they're just buying off-the-shelf servers from Supermicro/Dell (because Supermicro/Dell have a "vertical" 4U 40HDD server available for cheap now). Maybe Supermicro/Dell only uses Toshiba or something.

All in all: Backblaze doesn't care about reliability of drives. They kept a whole slew of Toshibas back when their failure rates were 20% or something. Backblaze's architecture literally doesn't care about the reliability of drives (see my previous post about tipping over racks). They care far more about how cheaply it is to load a rack of servers. Backblaze only publishes these hard drive statistics as a marketing effort: they know that by providing these statistics, people are more likely to buy their web-service.

 

[Steevo]

One or two people having drive failures from entirely different drives in entirely different environments VS literally thousands of drives in the same environment.

But somehow that data is wrong, it disagrees with my own circumstance and personal bias. Sure thing boss.

 

[zlobby]

While those statistics seem nice, they are flawed.

First, they don't really apply to the consumer, as they are using drives in environments they aren't meant to be used in. Also, what Backblaze considers a failure isn't what a failure would be in a single drive consumer application. Backblaze labels the drive as failed if the RAID(or whatever redundancy system they are using) kicks the drive out of the array. But when you are using consumer drives in multiple drive array applications, that can just happen even if the drive is perfectly fine. Sometimes it comes down to just the drive not responding to a request fast enough, which wouldn't be a problem in a single drive configuration but is in a multi-drive configuration.

Second, do you find it odd that they only use 2 WD models? In fact, I think for a while they had no WD models. You really have to ask yourself why a company that are running huge numbers of drives eventually just stopped using one brand for a while. And they answer this question, they had some WD model drives with 100% failure rates that they didn't even bother to include in the quarterly stat reports. They've said this on their blog before.



Yeah , I just put an 8TB for my games and minor other storage in my main PC a few months ago and I'm already down to less than 1TB free.

Say what you will, I'm not touching Seagate anytime soon. In my humble personal experience these are abysmal.

 

[newtekie1]

Backblaze doesn't have a redundancy system at the hardware level. Its their "web-software" that offers redundancy. Your data is stored in 20-different racks, in 20-different computers. At the OS-level, all of their drives are just JBODs (aka: Just a bunch of Drives). No RAID, no striping, no mirroring, no nothing. The web-software "manually" performs a Reed-Solomon 16-data+4-parity calculation.

That really doesn't have anything to do with how the drives are managed and treated as failed. Each "pod" that they use monitors the drives inside of it and has its own data redundancy system built in. Yes it is a software solution, not hardware, but so is unRAID or anything that uses ZFS. The software solutions still incorrectly label drives as failed.

Don't overthink it. There's lots of mundane reasons to cut out a manufacturer.

I like how you quoted just half a sentence pretending I didn't give the information as to why they cut out WD. They posted it themselves in a blog post, they had WD drive models that had 100% failure rates. There is no overthinking it, that is the reason, they said it themselves.

Say what you will, I'm not touching Seagate anytime soon. In my humble personal experience these are abysmal.

And I won't touch a WD Blue drive with a 10ft pole. Why? Because I do hundreds of failed hard drive replacements in customer's computers every year, and when a computer with a dead HDD comes in my shop it's almost always a dead WD Blue. But I won't touch a Seagate Barracuda drive either, because if it isn't a dead WD Blue, it's a dead Barracuda(or whatever Seagate was calling them at the time the computer was made).

The fact is, it's more about the model of the drive than the manufacturer. I mean, I won't buy a WD NAS drive because you never know if you are going to get a SMR drive. While Seagate never uses SMR in their NAS lines. But a WD Black for a single drive setup I'd still buy.

 

[jesdals]

It is not just my own personal experience, it well known that Seagate drives have a high failure rate.

Well that dosent say much unless you are a professional user. I have build and used Seagate since 1995 and only had one disk die. Still I am not going to say Seagate cant fail or is better than WD. But would I choose Seagate - proberly

 

[zlobby]

Oh, shite! Drive war! Drive war!