Western Digital planning 40 TB HDDs by 2025 with assistance of MAMR

[WMGroomAK]

Western Digital is trying to one up itself with news today of plans to begin production of MAMR (Microwave-Assisted Magnetic Recording) HDDs to ship in 2019 for Big Data Centers.  Long term, they are looking at being able to scale these to at least 40 TB drives by 2025 and offer over 4 Tb per a square inch.

http://www.businesswire.com/news/home/20171011005969/en/Western-Digital-Unveils-Next-Generation-Technology-Preserve-Access

Quote

At its “Innovating to Fuel the Next Decade of Big Data” event today, Western Digital Corp. (NASDAQ: WDC) announced a breakthrough innovation for delivering ultra-high capacity hard disk drives (HDDs) to meet the future demands of Big Data with proven data center-level reliability. The event, held at the company’s headquarters in Silicon Valley, included a demonstration of the world’s first microwave-assisted magnetic recording (MAMR) HDD and presentations from company executives and the inventor of MAMR technology, Professor Jimmy Zhu from Carnegie Mellon University. The company also showcased advancements in micro actuation and Damascene recording head technology. Western Digital expects to begin shipping ultra-high capacity MAMR HDDs in 2019 for use in data centers that support Big Data applications across a full range of industries.

...

MAMR is one of two energy-assisted technologies that Western Digital has been developing for years. The company recently innovated a breakthrough in material and process that provides the required reliable and predictable performance, as well as the manufacturability to accelerate areal density and cost improvements to an estimated average of 15 percent per year. Developments in the other energy-assisted technology, specifically, heat-assisted magnetic recording (HAMR), present new material science and reliability challenges that are not a factor in MAMR. Only MAMR demonstrates the reliability and cost profile that meets the demands of data center operators.

At the heart of the company’s innovation breakthrough is the “spin torque oscillator” used to generate a microwave field that increases the ability to record data at ultra-high density without sacrificing reliability. Western Digital’s innovative MAMR technology is expected to offer over 4 terabits-per-square-inch over time. With sustained improvements in recording density, MAMR promises to enable hard drives with 40TB of capacity and beyond by 2025, and continued expansion beyond that timeframe

So, while you still will be limited on read write speeds, this will probably help to keep spinning rust around for another decade or so as it'll drastically increase the storage density for large data centers...  Should be interesting to see how well Solid State storage also scales going forward and look at how cost per TB begins to work out...

 

Hot Hardware: https://hothardware.com/news/western-digital-mamr-hard-drive-density-to-scale-north-of-40tb

 

EDIT:  Here is the PC Per article, which has a really good breakdown on the advantages and differences of HAMR vs MAMR technology...

 

https://www.pcper.com/news/Storage/Western-Digital-MAMR-Tech-Pushes-Future-HDDs-Beyond-40TB

 

The Anandtech Article is fairly decent as well... Especially in the presentation of the advancement of HDD technology over the last 15 to 20 years.

 

https://www.anandtech.com/show/11925/western-digital-stuns-storage-industry-with-mamr-breakthrough-for-nextgen-hdds

 

 

Edited October 13, 2017 by WMGroomAK
Adding Anandtech Article

[8uhbbhu8]

Damn.... This is actually pretty amazing that they can even think of getting that much data on a drive.

 

 

Wonder what effect it will have on main stream smaller storage

[ScratchCat]

Spoiler

We were supposed to have this technology in 2018, not 2025.

It seems HDD technology is reaching a limit and cannot keep up with the development rate of SSDs anymore. Once SSDs approach HDDs in cost per GB it will be over for HDDs, although this will still be a few years away. Cost per GB of SSDs has quartered since 2012 while cost of an HDD probably has gone down 20% per GB. HDDs will be replaced soon by SSDs in any system not used in archival of large amounts of infrequently used data.

[Guest]

By 2025 i was expecting 500-600TB HDDs xd....

[8uhbbhu8]

2 minutes ago, ScratchCat said:

  Hide contents

We were supposed to have this technology in 2018, not 2025.

It seems HDD technology is reaching a limit and cannot keep up with the development rate of SSDs anymore. Once SSDs approach HDDs in cost per GB it will be over for HDDs, although this will still be a few years away. Cost per GB of SSDs has quartered since 2012 while cost of an HDD probably has gone down 20% per GB. HDDs will be replaced soon by SSDs in any system not used in archival of large amounts of infrequently used data.

I think HDDs will still stick around for a long time due to the negatives of flash storage compared to disk storage.

[WMGroomAK]

4 minutes ago, ScratchCat said:

We were supposed to have this technology in 2018, not 2025.

HAMR drives are projected to launch in 2018, although the drives that WD is talking about now use MAMR so they can operate with less heat.  I think Seagate is planning to launch HAMR in 2018 and WD is talking about MAMR in 2019 with projections of reaching the 4 Tb/in² mark by 2025.  Going by the chart, the MAMR drives may only be a couple of years behind schedule of the HAMR drives...  Long term, I do think that these will only be used in Archival storage, especially if Solid State memory can increase it's storage density on a similar scale.

[Misanthrope]

I'd be surprised if SSDs haven't reached 40tb by 2025

[8uhbbhu8]

2 minutes ago, Misanthrope said:

I'd be surprised if SSDs haven't reached 40tb by 2025

Seagate already had a 60TB Last Year!

[S w a t s o n]

So basically we're gonna see these HDD's replace tape drives for archival while SSD's replace HDD's for live storage. Sweet. My next build is planned to be SSD only anyways.

[TrigrH]

7 hours ago, Swatson said:

So basically we're gonna see these HDD's replace tape drives for archival while SSD's replace HDD's for live storage. Sweet. My next build is planned to be SSD only anyways.

I've been SSD only for over 2 years now, worth every cent.

[samcool55]

9 hours ago, ScratchCat said:

  Reveal hidden contents

We were supposed to have this technology in 2018, not 2025.

It seems HDD technology is reaching a limit and cannot keep up with the development rate of SSDs anymore. Once SSDs approach HDDs in cost per GB it will be over for HDDs, although this will still be a few years away. Cost per GB of SSDs has quartered since 2012 while cost of an HDD probably has gone down 20% per GB. HDDs will be replaced soon by SSDs in any system not used in archival of large amounts of infrequently used data.

SSD pricing went up since the beginning of this year...

We are actually right now heading the wrong direction when it comes down to SSD pricing.

The gap is growing instead of getting smaller 

 

Anyway, i"m wondering how fast those drives can read/write.

If they are stuck at like 100MB/s it would take like 3-4 days to fill up the disk completely, just stupid...

[ScratchCat]

9 hours ago, 8uhbbhu8 said:

I think HDDs will still stick around for a long time due to the negatives of flash storage compared to disk storage.

Apart from cost there are few to no negatives for flash storage , the limited read write cycles are countered by non correctable errors in HDDs.

 

9 hours ago, WMGroomAK said:

HAMR drives are projected to launch in 2018, although the drives that WD is talking about now use MAMR so they can operate with less heat.  I think Seagate is planning to launch HAMR in 2018 and WD is talking about MAMR in 2019 with projections of reaching the 4 Tb/in² mark by 2025.  Going by the chart, the MAMR drives may only be a couple of years behind schedule of the HAMR drives...  Long term, I do think that these will only be used in Archival storage, especially if Solid State memory can increase it's storage density on a similar scale.

I meant that level of technology , specifically the 4TB/in2. According to that graph we should have obtained this near 2018-2019 , not 2025. 

 

17 minutes ago, samcool55 said:

SSD pricing went up since the beginning of this year...

We are actually right now heading the wrong direction when it comes down to SSD pricing.

The gap is growing instead of getting smaller 

 

Anyway, i"m wondering how fast those drives can read/write.

If they are stuck at like 100MB/s it would take like 3-4 days to fill up the disk completely, just stupid...

I'm just getting on the price following to normal trend once the manufacturers increase thier production levels and the price returns to it's normal rate of change.

[leadeater]

10 hours ago, Swatson said:

So basically we're gonna see these HDD's replace tape drives for archival while SSD's replace HDD's for live storage. Sweet. My next build is planned to be SSD only anyways.

Nope, because LTO smashes HDDs for that type of usage.

 

 

LTO-8 is coming out very soon at 12.8TB uncompressed and 32TB compressed with write performance of 360 MB/s uncompressed and 750 MB/s, an HDD just won't be able to compete with that let alone LTO-10 which should be out in 2024 (each generation is 3 years ish).

 

Edit:

By soon I mean LTO-8 is Q4 2017.

http://www.storagereview.com/ibm_releases_30tb_lto8_tape_drive

[Dylanc1500]

11 hours ago, leadeater said:

Nope, because LTO smashes HDDs for that type of usage.

 

 

LTO-8 is coming out very soon at 12.8TB uncompressed and 32TB compressed with write performance of 360 MB/s uncompressed and 750 MB/s, an HDD just won't be able to compete with that let alone LTO-10 which should be out in 2024 (each generation is 3 years ish).

 

Edit:

By soon I mean LTO-8 is Q4 2017.

http://www.storagereview.com/ibm_releases_30tb_lto8_tape_drive

Don't tease me.

 

In reality I don't think tape or hard disks will go anywhere for a very long time. All that is happening is they are being more cemented into their best use case for the market.

[paddy-stone]

I've been SSD only for quite a while now on my main rig, everything not conducive to operating the rig gets pushed to the NAS. I can't believe I didn't do it sooner TBH.

 

On topic, I can't remember the price of HDDs falling for around a few years now, at least in the 3TB and under segment.

It's good seeing these new technologies emerge, but if they have to have a minimal HDD size of say 20TB to make the drives worth producing, I think it'll be beyond my scope of caring for at least another 10 years after they release. They'll just be too expensive for my taste until my level of storage comes near to the point where I have to start replacing the drives in the array.

[TVwazhere]

3 hours ago, leadeater said:

Nope, because LTO smashes HDDs for that type of usage.

 

 

LTO-8 is coming out very soon at 12.8TB uncompressed and 32TB compressed with write performance of 360 MB/s uncompressed and 750 MB/s, an HDD just won't be able to compete with that let alone LTO-10 which should be out in 2024 (each generation is 3 years ish).

 

Edit:

By soon I mean LTO-8 is Q4 2017.

http://www.storagereview.com/ibm_releases_30tb_lto8_tape_drive

Also, dont tape drive literally last forever?

 

[leadeater]

3 minutes ago, TVwazhere said:

Also, dont tape drive literally last forever?

Tapes hate moisture so you need to store them properly, other than that they'll store data longer than you'll care to get the data back off them.

[WMGroomAK]

9 hours ago, ScratchCat said:

I meant that level of technology , specifically the 4TB/in2. According to that graph we should have obtained this near 2018-2019 , not 2025. 

If I'm reading that graph correctly, then it shows HAMR+ as reaching 4 TB/in² by late 2020 or 2021, not 2018 or 2019.  The Y-Axis is logarithmic on it...  By 2018-2019, it would appear to be between 1.8 and 3 Tb/in².  It's really not that great of a graph as the arrows are well to wide to be meaningful.

[S w a t s o n]

7 hours ago, leadeater said:

snip

While this is true, tape drives are not without inherent disadvantages. Being read from sequentially for instance, esp on a large drive like this could be a pain if you only want to access the latter portion of the drive. I also believe the ubiquity of HDD technology will lead these drives to eventually crush tape drives in terms of cost and TCO. Tape drives aren't cheap. It's another machine in the building to have to support. though the actual catridges themselves are cheap.

[Doobeedoo]

Definitely great storage capaciy, tough limited in speed due to being mechanical, but will have it's use no doubt. Like on consumer side, storing 8K footage and such, though installing a very large game will take time. Like said, while capacity increases and demands for such, speeds stay the same. Price is also a thing. 

For SSDs it will be very interesting once they close in and approach to HDD prices. Still are few times more expensive TB/price. 

 

[Dylanc1500]

17 minutes ago, Swatson said:

While this is true, tape drives are not without inherent disadvantages. Being read from sequentially for instance, esp on a large drive like this could be a pain. I also believe the ubiquity of HDD technology will lead these drives to eventually crush tape drives in terms of cost and TCO. Tape drives aren't cheap. It's another machine in the building to have to support. though the actual catridges themselves are cheap.

This is why they are meant for hard archiving. These aren't for data you will frequently access. Which is why @leadeater said you'll be able to store the data longer than you care to get the data back off.

[leadeater]

2 hours ago, Swatson said:

While this is true, tape drives are not without inherent disadvantages. Being read from sequentially for instance, esp on a large drive like this could be a pain if you only want to access the latter portion of the drive. I also believe the ubiquity of HDD technology will lead these drives to eventually crush tape drives in terms of cost and TCO. Tape drives aren't cheap. It's another machine in the building to have to support. though the actual catridges themselves are cheap.

HDD have their purpose but so do tapes so it's hard for either one of them to replace the other.

 

An LTO-7 drive is only $7k USD real price, LTO-7 tape cost depends on how many you buy at once but you can get them between $50-$100 each.

 

What makes tapes really win out currently though is the cost to performance ratio, when you have a very large amount of data you need to archive or long term backup you'll get that on to tape much quicker than to HDD. There is a reason all disk based backup systems today are incremental forever and never do a full backup after the first, the time it would take.

 

We do disk backups to a 4 node Netapp 3220 (being replaced soon) with 288 disks and the highest sequential sustained throughput we can get out of that is 1500 MB/s, we would have problems getting enough throughput to the tape drives even to a modest 4 drive 80 tape library.

 

Essentially HDDs will never replace tapes for real long term archiving and backups for very large data sets while their write performance per drive is below 200MB/s and the cost per disk is as high as they are now.

 

If you go back and watch the video Linus did with the University of Canada with all their super fast disk storage with scale out and up capability they still put in a large robotic tape library, they wouldn't sink $250k-500k ish in 2017 in to tapes if it wasn't the best solution for that use case now and in to the future.

[MyName13]

I really hope SSDs will replace HDDs by then on the consumer market, it's ridiculous how we still have mechanical hardware in our computers.

[S w a t s o n]

5 hours ago, leadeater said:

University of Canada

Wait a minute, do we have a federal university I'm not aware of?

 

But for seriously, while I do see your point, these new drives are not available yet. You can't say the University that he worked with bought tape drives because they are the better option when they didn't even know of these drives and even if they did know, they wont be available for some time. Long enough that maybe they couldn't wait. I do see what you mean though.

[leadeater]

9 hours ago, Swatson said:

Wait a minute, do we have a federal university I'm not aware of?

That's my "crap forgot which one but you'll know which one I'm talking about", lol.

 

9 hours ago, Swatson said:

But for seriously, while I do see your point, these new drives are not available yet. You can't say the University that he worked with bought tape drives because they are the better option when they didn't even know of these drives and even if they did know, they wont be available for some time. Long enough that maybe they couldn't wait. I do see what you mean though.

Both technologies are advancing with reasonable projections, tape does have a bit more clear projections than HDDs, so it's not going to be a case of one surpassing the other unless that changes. Either one has to have a big breakthrough or one has to hit a significant issue preventing development. 

 

Anyway based on these drives announced just now which might exist in 2025 tape is still the better option. LTO-10 in 2024 will be able to store 48TB/120TB projected which is up to 3 times above this projected 40TB and the tape will be able to write the data much much faster. That University can also just change out the tape drives in the library as newer better ones come out for little cost and effort and cycle new tapes in as per normal operation.

 

I have to deal with both disk backup and tape backups daily so I know which is good for what quite well, we have 2x 5PB of deduplicated data on disk backup and 2500 LTO-5 tapes (we need to upgrade to LTO-7). There are things I'd never do with tape backup and things I'd never do with disk backup.