Stories
Slash Boxes
Comments

SoylentNews is people

SoylentNews is powered by your submissions, so send in your scoop. Only 9 submissions in the queue.
posted by martyb on Thursday November 09 2017, @06:32AM   Printer-friendly
from the hammer-hamer-mamer-mammer dept.

Western Digital recently announced plans to use Microwave Assisted Magnetic Recording (MAMR) to build its next generation of hard disk drives instead of Heat Assisted Magnetic Recording (HAMR). WD promised that initial drives will ship in 2019, with 40 terabyte drives available by 2025.

In response, Seagate has reiterated its plans to produce HAMR hard disk drives in the near future. The company says that its first HAMR drives will ship around 2018-2019 (40,000 have already been built and are being tested by leading customers), at capacities of 16 TB or more. From there, Seagate expects to develop drives storing around 50 TB "early next decade", and eventually drives with capacities of up to 100 TB by combining HAMR with bit-patterned media and two-dimensional magnetic recording (PDF):

HDD technology has become somewhat boring. Innovation has slowed, but that's largely because we've reached the limits of PMR (Perpendicular Magnetic Recording), which is the key underlying HDD recording technology. Over the last two years, we've seen a few interesting new technologies that let us cram more bits into the same old 3.5" HDD, such as SMR (Shingled Magnetic Recording). Unfortunately, the new tech comes with slower performance and often requires radical system changes if you want to unlock the full performance. That isn't worth the small capacity improvement unless you're deploying tens of thousands of HDDs.

[...] WD's MAMR relies largely upon proven technologies, which is a plus, but Seagate claimed that it's already producing the more exotic HAMR drives on the same production lines as its existing PMR-based drives. It also said that it has already built a strong supply chain for the new materials.

Both WD and Seagate have solid arguments for their chosen technologies, but the market will determine the winner. Both technologies will undoubtedly provide similar characteristics to today's HDDs, such as endurance, reliability, performance, and power specifications, so cost will be the true differentiator. As always, cheap and good enough will win. The HDD industry settled on PMR recording in 2005, and all three big vendors continue to use the same underlying technology. The move to two different technologies should make for a more exciting HDD future. Seagate plans to provide an update on its progress in early 2018.

Previously: AnandTech Interview With Seagate's CTO: New HDD Technologies Coming
Seagate HAMR Hard Drives Coming in a Year and a Half
Glass Substrate Could Enable Hard Drives With 12 Platters


Original Submission

 
This discussion has been archived. No new comments can be posted.
Display Options Threshold/Breakthrough Mark All as Read Mark All as Unread
The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
(1)
  • (Score: 0) by Anonymous Coward on Thursday November 09 2017, @06:45AM (8 children)

    by Anonymous Coward on Thursday November 09 2017, @06:45AM (#594474)

    When I backup my computer I simply rsync all its files. I keep meaning to organize all my backups, de-duplicate them, etc,,, Instead when the drive fills up I buy a new one, copy everything over, and then keep syncing. Sounds like I'll never be pressured to clean up my digital hoards.

    • (Score: 3, Interesting) by takyon on Thursday November 09 2017, @07:17AM (4 children)

      by takyon (881) <reversethis-{gro ... s} {ta} {noykat}> on Thursday November 09 2017, @07:17AM (#594484) Journal

      Write speeds are not scaling to capacity at all, so it's going to take longer and longer to move the growing pile of files, unless your hoard isn't growing that much each time and is filling a smaller % of each new drive.

      Top SSD capacity has already blown past what HDDs can achieve in the next ten years, and will probably hit 1,000 TB in a 3.5" enclosure before long. But if we believe WD, Seagate, and analysts, HDDs will continue to remain competitive on $/TB.

      --
      [SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]
      • (Score: 0) by Anonymous Coward on Thursday November 09 2017, @08:11AM (1 child)

        by Anonymous Coward on Thursday November 09 2017, @08:11AM (#594505)

        People have been saying that for years, but every time I upgrade my system, SSD sizes seem to be several generations behind.

        I recently upgraded from a 1TB drive to a 4 TB drive, and while I was comparing prices, I checked SSDs also. I could get 250GB, maybe even 500, but anything above that I would have to pick a server drive adding at least one zero - and I still don't even think they had anything in the 4TB range.

        • (Score: 4, Informative) by takyon on Thursday November 09 2017, @08:28AM

          by takyon (881) <reversethis-{gro ... s} {ta} {noykat}> on Thursday November 09 2017, @08:28AM (#594516) Journal

          The products in the summary will almost certainly be sold to datacenters first, like today's top-of-the-line helium-filled 14 TB hard drives.

          A 15 TB SSD [computerworld.com] has been shipping for over a year and a half now. It was originally priced at $10,000.

          32 TB [zdnet.com] and 60 TB [arstechnica.com] SSDs exist and are probably around somewhere. 100 TB [theregister.co.uk] and 128 TB [soylentnews.org] SSDs have been announced. So SSDs have blown past hard drives in terms of capacity and by the time HDDs reach 50 TB, SSDs could be past 1,000 TB. QLC NAND is replacing TLC NAND, layers are increasing, and dies are getting stacked. All at once. The petabyte barrier will be broken, even in the 2.5" format.

          The largest consumer-oriented HDDs on the market are 12 TB [soylentnews.org].

          Consumers can buy Samsung's 4 TB 850 EVO for about [amazon.com] $1,550 [newegg.com]. With 1 TB SSDs [newegg.com] around $270, that purchase would not make a lot of sense for most people. 8 TB consumer SSDs are probably not far behind, but it could take 1-2 years and a new generation of 3D QLC NAND before anyone considers launching a consumer-oriented 8 TB SSD.

          --
          [SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]
      • (Score: 2) by richtopia on Thursday November 09 2017, @04:50PM (1 child)

        by richtopia (3160) on Thursday November 09 2017, @04:50PM (#594695) Homepage Journal

        When Seagate's CEO Stephen Luczo announced the intention to sell HAMR drives in 2018, he also recognized that competing with SSDs are a losing battle:

        “By this time next year, we anticipate less than 10% of our HDD technology portfolio will be exposed to competing flash devices,” said Mr. Luczo.

        Yes, times have been and will be tough for the spinning disc manufacturers. However they realize they have to adapt to the market going forward and it sounds like they are making decisions to ensure they provide a unique, competitive solution.

    • (Score: 0) by Anonymous Coward on Thursday November 09 2017, @07:18AM (1 child)

      by Anonymous Coward on Thursday November 09 2017, @07:18AM (#594485)

      one day i'm leaving midget porn on an unencrypted isolinear chip so i can troll my grandchildren when they're looting my treasures - Surprise, Motherfuckers!

    • (Score: 3, Informative) by Azuma Hazuki on Thursday November 09 2017, @06:28PM

      by Azuma Hazuki (5086) on Thursday November 09 2017, @06:28PM (#594745) Journal

      You can use a small Bash script and the --delete and --link-dest switches in rsync to do incremental backups. I use a ~10 line script to do this over SSH for some commercial clients on their LANs, after having the machines exchange SSH keys. Basically, the script makes a temp directory, copies everything into it, datestamps it, and then repeats the process, with each subsequent run comparing what's in the temp dir (--link-dest switch points at it) with what's to be backed up. This forces rsync to make hard links rather than full copies, so subsequent runs take up virtually no space for already-existing stuff.

      --
      I am "that girl" your mother warned you about...
  • (Score: 2) by takyon on Thursday November 09 2017, @07:12AM

    by takyon (881) <reversethis-{gro ... s} {ta} {noykat}> on Thursday November 09 2017, @07:12AM (#594481) Journal

    Tom's Hardware massaged some weird numbers out of Seagate's blog post, like 20+ TB in 2019 and 40+ TB by 2023. I have gone with what you see in the headline instead.

    --
    [SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]
  • (Score: 2) by ledow on Thursday November 09 2017, @08:07AM (2 children)

    by ledow (5567) on Thursday November 09 2017, @08:07AM (#594501) Homepage

    Cool.

    You do that Seagate. It's where you belong.

    Meanwhile, the first company to produce a decent-size, decent-price, SSD will get LOTS of my custom (doesn't even need to be a hugely fast NVMe thing... just SATA speed SSDs are more than capable of doubling the speed of a machine).

    I honestly don't get why we're even bothering with hard drives any more.

  • (Score: 1, Interesting) by Anonymous Coward on Thursday November 09 2017, @08:08AM (8 children)

    by Anonymous Coward on Thursday November 09 2017, @08:08AM (#594503)

    When I was in school, somewhere between 1992 and 1995, I was "stalking" our sysadmin, and at one point a technician from the schools IT supplier were upgrading the Unix system. He talked that the next generation of hard drives would not have a moving read/write head, but one head per track.

    That would make hard drives almost as fast as SSDs, theoretically without reducing the capacity to SSD levels (of course that would require being able to make the read/write electronics small enough).

    25 years later, we still haven't seen these hard drives.

    Meanwhile hard drive sizes are going up, along with the size of the data they contain (compare a 4K bluray rip with an animated GIF from the 1990'es), but the speeds are not increasing nearly as much.

    And SSDs? I'm still waiting for one that can replace a hard drive and not only act as a cache for the most used files. My OS already uses RAM as a disk cache, and that's a lot faster than an SSD. And by replace, I mean TB in size, not GB, and costing less than twice as much as a HDD.

    • (Score: 0) by Anonymous Coward on Thursday November 09 2017, @09:15AM (1 child)

      by Anonymous Coward on Thursday November 09 2017, @09:15AM (#594542)

      have you considered the possibility that the technician was making fun of you?

      • (Score: 0) by Anonymous Coward on Thursday November 09 2017, @12:56PM

        by Anonymous Coward on Thursday November 09 2017, @12:56PM (#594575)

        He was not talking to me, he was talking to our sysadmin.

    • (Score: 4, Informative) by TheRaven on Thursday November 09 2017, @09:49AM (5 children)

      by TheRaven (270) on Thursday November 09 2017, @09:49AM (#594548) Journal

      That's been proposed on and off since winchester disks were first created. The problem with it is that the read head is a lot wider than the track, so you end up having to stagger them in a spiral to pack them in tightly. You then end up with a hard disk that's almost completely covered in read heads, which means that you can't spin it very fast if you don't want it to melt its way out of the bottom of the computer, at which point you're likely to lose more on sequential read speeds than you gain on random read speeds. It's an engineering problem that looks simple at first glance and ends up being a lot more complex.

      I think a few disk models have had two or three read heads. This can speed things up a lot, because the speed random accesses is dominated by either the movement time (which is smaller if you move less far) and the settling time, which is much smaller if the head moves more slowly, and you can move the head a lot more slowly if it doesn't have to move as far. As I recall, these largely went out of favour as 3.5" and 2.5" disks became common, because the head-movement distance on a 2.5" disk is already very small (especially compared to the 5.25" disks that were common 20 years go).

      --
      sudo mod me up
      • (Score: 2) by letssee on Thursday November 09 2017, @02:30PM (4 children)

        by letssee (2537) on Thursday November 09 2017, @02:30PM (#594598)

        I'd imagine that copying data from one place of the drive to another would always be a lot faster when having at least 2 heads. I'm actually suprised we haven't seen the multiheaded disks very much.

        • (Score: 2) by TheRaven on Thursday November 09 2017, @06:08PM (2 children)

          by TheRaven (270) on Thursday November 09 2017, @06:08PM (#594737) Journal
          The models I've seen with two heads had them at fixed offsets, so moving one moved the other by the same amount (otherwise you need an extra arm, which adds a lot of complexity and needs a lot of space). Having two independent heads, one for reads and one for rights, and ensuring that they didn't interfere would probably be quite difficult.
          --
          sudo mod me up
          • (Score: 2) by letssee on Thursday November 16 2017, @09:28AM (1 child)

            by letssee (2537) on Thursday November 16 2017, @09:28AM (#597623)

            I'd think 2 arms on opposite sides of the drive should work. But you are probably right, because if it was easy we would have seen them more often

            • (Score: 2) by TheRaven on Thursday November 16 2017, @10:35AM

              by TheRaven (270) on Thursday November 16 2017, @10:35AM (#597633) Journal
              Yes, I can see the potential speedup from doing that, but if you take a hard disk apart there really isn't much space for extra actuators to move another head. You could probably do it on 5.25" disks quite easily these days, especially if you used a 3.5" disk and put the motors for moving the heads on the outside, but the market for 5.25" disks is pretty small and I doubt anyone would pay enough for them for them to be worthwhile. The problem is that you wouldn't be doing much about worst-case latency (if neither head is in the right place, you'll need to move one, and even if you need to move it half the distance of a conventional drive it's still a few ms) and you'd only cut average-case latency in half. If anyone has a workload that's latency-critical enough to want to do this, they'd be better off buying SSDs with latency around 10-100 times lower.
              --
              sudo mod me up
        • (Score: 2) by bob_super on Thursday November 09 2017, @07:03PM

          by bob_super (1357) on Thursday November 09 2017, @07:03PM (#594768)

          It would make sense to have two to four sets of heads per platter side, but that would make the drives a lot more expensive (closer to SSD prices) and drop the MTBF significantly (for full operation).
          I'm not sure how the mechanical aspects of moving three heads while one is writing would work. Some pretty tricky vibration to account for.

  • (Score: 2) by DannyB on Thursday November 09 2017, @04:57PM (5 children)

    by DannyB (5839) Subscriber Badge on Thursday November 09 2017, @04:57PM (#594696) Journal

    Since hard drives are getting so big, it seems the time has come to start putting Intel-like "management engines" within hard drives. hard drives could have a secondary network connector that bypasses the primary interface to the computer.

    Such large drives beg for an industrial strength filesystem. Or a LOT of partitions.

    If the management engine provides "read only" access to the drive, then the mother ship TLAs would be able to determine what the file system and/or partition layout is. But for write access, such as to plant evidence, it might be good if the management engine understood the partition and filesystem layout of the drive. That way a lot less network traffic is needed for perusing the drive for crypto keys or when planting evidence onto the drive, but not through the main CPU.

    You think I'm kidding, right?

    --
    When trying to solve a problem don't ask who suffers from the problem, ask who profits from the problem.
    • (Score: 2) by Azuma Hazuki on Thursday November 09 2017, @06:30PM (2 children)

      by Azuma Hazuki (5086) on Thursday November 09 2017, @06:30PM (#594748) Journal

      I hope to sweet baby Cthulhu you are. The security and civil rights implications are horrifying...

      --
      I am "that girl" your mother warned you about...
      • (Score: 2) by DannyB on Thursday November 09 2017, @09:07PM (1 child)

        by DannyB (5839) Subscriber Badge on Thursday November 09 2017, @09:07PM (#594834) Journal

        Security? Civil Rights? You use such antiquated and unfamiliar terms my friend.

        --
        When trying to solve a problem don't ask who suffers from the problem, ask who profits from the problem.
        • (Score: 2, Troll) by Azuma Hazuki on Saturday November 11 2017, @05:37AM

          by Azuma Hazuki (5086) on Saturday November 11 2017, @05:37AM (#595504) Journal

          Mama always did say I was an old soul...

          --
          I am "that girl" your mother warned you about...
    • (Score: 2) by letssee on Thursday November 16 2017, @09:33AM (1 child)

      by letssee (2537) on Thursday November 16 2017, @09:33AM (#597625)
(1)