Stories
Slash Boxes
Comments

SoylentNews is people

SoylentNews is powered by your submissions, so send in your scoop. Only 17 submissions in the queue.
posted by hubie on Friday May 09, @04:33PM   Printer-friendly
from the "might-be" dept.

Arthur T Knackerbracket has processed the following story:

Cerabyte recently conducted an experiment that seemed more like a culinary exercise than a technology showcase. The German storage startup plunged a sliver of its archival glass storage into a kettle of boiling salt water, then roasted it in a pizza oven.

Despite enduring temperatures of 100°C in the kettle and 250°C in the oven, the storage medium emerged unscathed, with its data fully intact. This experiment – along with a similar live demonstration at the Open Compute Project Summit in Dublin – was not just a spectacle. It was Cerabyte's way of proving a bold claim: its storage media can withstand conditions that would destroy conventional data storage.

Founded in 2022, Cerabyte is on a mission to upend the world of digital archiving. The company's technology relies on an ultra-thin ceramic layer – just 50 to 100 atoms thick – applied to a glass substrate.

Using femtosecond lasers, data is etched into the ceramic in nanoscale holes. Each 9 cm² chip can store up to 1 GB of information per side, written at a rate of two million bits per laser pulse. Cerabyte claims the result is a medium as durable as ancient hieroglyphs, with a projected lifespan of 5,000 years or more.

The durability of glass is well known. Its resistance to aging, fire, water, radiation, and even electromagnetic pulses makes it a natural candidate for "cold storage." Cerabyte's tests – including boiling the media in salt water for days (long enough to corrode the kettle itself) and baking it at high heat – were designed to underscore this resilience.

While the company has not disclosed how the ceramic layer or its bond to the glass would fare under physical shock, the media's resistance to environmental hazards is clear.

Cerabyte's ambitions extend beyond durability. The startup aims to reduce the cost of archival storage to less than $1 per terabyte by 2030 – a target that could transform the economics of long-term data retention.

[...] Unlike other archival methods – magnetic tape, hard drives, or even optical discs, all of which degrade over decades – Cerabyte's ceramic-on-glass approach promises to eliminate the need for regular data migration or energy-hungry maintenance.


Original Submission

This discussion was created by hubie (1068) for logged-in users only. Log in and try again!
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: 3, Interesting) by Username on Friday May 09, @04:46PM (10 children)

    by Username (4557) on Friday May 09, @04:46PM (#1403199)

    A dvd/bluray is 4.7", this is 3.5". Blu-ray is 25gb to 100gb per disc, this is 1gb.

    They're differentiating this from optical media, so a laser isn't going to read the data? It's a hdd platter?

    I bet if I scratch that 100 atom thick ceramic with my fingernail it will stop working.

    • (Score: 5, Interesting) by VLM on Friday May 09, @05:39PM (3 children)

      by VLM (445) Subscriber Badge on Friday May 09, @05:39PM (#1403204)

      I always wanted an optical media that used an encoding as strong as par files.

      https://en.wikipedia.org/wiki/Parchive [wikipedia.org]

      I had a lot of experience with par files back in the usenet decades.

      I'd be happy with an entire optical disk that can only store a single couple of K RSA key thats still readable as long as 1% of the media isn't scratched up.

      I'm surprised nothing like this ever hit the market.

      • (Score: 2, Insightful) by anubi on Saturday May 10, @09:32AM (2 children)

        by anubi (2828) on Saturday May 10, @09:32AM (#1403278) Journal

        Sounds similar to the error correction on CDRom and QR coding to make data read back more robust with damaged media.

        For many things, I would gladly trade off file size for robust read and repair, but this is easier said than done when defense against deliberate targeted tampering is considered.

        --
        "Prove all things; hold fast that which is good." [KJV: I Thessalonians 5:21]
        • (Score: 4, Insightful) by VLM on Saturday May 10, @03:08PM (1 child)

          by VLM (445) Subscriber Badge on Saturday May 10, @03:08PM (#1403302)

          Similar but I want way more. Like if the iso image you're burning is only 1% of the disk, instead of quickly producing a 99% empty disk you get an enormous amount of error detection and correction such that I can recover the data if only 1% of the disk is undamaged.

          • (Score: 1) by anubi on Saturday May 10, @06:13PM

            by anubi (2828) on Saturday May 10, @06:13PM (#1403321) Journal

            Quite insightful...especially given media like CDROMS are of a standardized capacity, and cost the same to reproduce whether they are filled to capacity or not.

            I have many CDROMS with only a few MB on them.

            I like that idea...an ".iso burner" that uses all available storage for redundant error recovery...so things like my collection of schematics/PCB layout/parts for a project ( maybe 100 MB ) can be archived on a 500 MB ROM, 100% fill, optimized for error tolerance, as part of "closing" a ROM burn image.

            If one wants the minimal file size load, one can always use the archival read to de-expand the archive to the working media...or re-expand the working file set back into another maximal-robustness archival media fill.

            --
            "Prove all things; hold fast that which is good." [KJV: I Thessalonians 5:21]
    • (Score: 4, Insightful) by Captival on Friday May 09, @09:06PM (5 children)

      by Captival (6866) on Friday May 09, @09:06PM (#1403227)

      What physical media can you not destroy? All that matters is that it beats current write media which still have a tendency to die of old age even if you keep them pristine.

      • (Score: 2) by aafcac on Friday May 09, @10:23PM (4 children)

        by aafcac (17646) on Friday May 09, @10:23PM (#1403240)

        Everything can be destroyed, but I think punched aluminum is probably one of the most durable options. Once it oxidizes, it's pretty tough and hard to accidentally destroy.

        • (Score: 1) by anubi on Saturday May 10, @09:39AM

          by anubi (2828) on Saturday May 10, @09:39AM (#1403280) Journal

          I understand even those copper scrolls didn't all that well.

          ( I have seen heavy copper automotive wiring that didn't age well at all...in less than 50 years . Corrosion and work hardening. It became a green colored powder. )

          --
          "Prove all things; hold fast that which is good." [KJV: I Thessalonians 5:21]
        • (Score: 2) by DannyB on Monday May 12, @02:13PM (2 children)

          by DannyB (5839) Subscriber Badge on Monday May 12, @02:13PM (#1403555) Journal

          aluminum . . . Once it oxidizes

          Just mix with iron oxide (aka common "rust") and you have Thermite.

          Ignite thermite and it will burn through almost anything and keep going until the reactants are consumed.

          --
          The only way to stop a bad guy with a can opener is a good guy with a can opener.
          • (Score: 2) by aafcac on Monday May 12, @04:19PM (1 child)

            by aafcac (17646) on Monday May 12, @04:19PM (#1403571)

            If we're intentionally adding other things to the mix, then why not just run the things over with a steamroller?

            • (Score: 3, Funny) by DannyB on Tuesday May 13, @01:25PM

              by DannyB (5839) Subscriber Badge on Tuesday May 13, @01:25PM (#1403673) Journal

              It might be unintentional that rust gets introduced to the storage medium forming thermite.

              This could be avoided with C / C++.

              --
              The only way to stop a bad guy with a can opener is a good guy with a can opener.
  • (Score: 2) by Dr Spin on Friday May 09, @05:10PM (2 children)

    by Dr Spin (5239) on Friday May 09, @05:10PM (#1403200)

    ... pigs might fly.

    --
    Warning: Opening your mouth may invalidate your brain!
  • (Score: 2) by looorg on Friday May 09, @05:13PM (2 children)

    by looorg (578) on Friday May 09, @05:13PM (#1403201)

    then roasted it in a pizza oven. ... 250°C in the oven,

    That is quite cold for a pizza oven. It should be almost double that. I can get 250 in the standard oven I have in my kitchen.

    Still it's glass. It's fairly brittle. I didn't really read or look into it to much but I assume the glass discs are used with some kind of caddy system like CD-roms and disks of ye' olden times?

  • (Score: 5, Insightful) by VLM on Friday May 09, @05:34PM (6 children)

    by VLM (445) Subscriber Badge on Friday May 09, @05:34PM (#1403202)

    Cerabyte's ceramic-on-glass approach promises to eliminate the need for regular data migration or energy-hungry maintenance.

    Maint will never end, you have to keep transferring data every couple years as tech changes.

    One thing the journalist and marketing filter missed will be no weird HVAC for climate controlled storage.

    The most tightly controlled room I have ever been in was a very large tape storage room at a major financial services institution. Stuff that's not important enough to spend money on putting in the tape-robot for rapid access gets put on special shelves in a special room which I'm sure is terribly expensive although probably pretty cheap per petabyte compared to the competition. You can't remove this stuff from the building under NDA etc so cloud is out of the question. The kind of place that has a security guard but he's not armed (well, maybe in 2025 he is armed, I haven't been there in awhile)

    One funny probably non-NDA anecdote is I worked there both pre- and post- tape robot and you'd think the tape robot would result in the tape shuffler apprentices all getting fired; what it resulted in was storage being cheaper, SEC regulations tightening, and now the same number of tape shuffler apprentices worked transferring tape between ultra-deep cold storage and the tape robot. Sure, department wide the cost per petabyte dropped quite a bit, its just business/regulation demands magically increase such that they had about the same number of employees before and after. There's still companies where the fastest bandwidth for data transfer is shipping media. Also there's a chain of custody thing WRT physical media, in theory if something disappeared or copies mysteriously appeared on the internet they could trace to individual human couriers. So you'd have crates of tape media show up that need insertion/removal/copying into the tape robot. The work never seems to end... I guess they need something for the apprentices to do while learning mainframe stuff. I don't know what the noobs would do without tape drives.

    Glass is notoriously weak in alkaline solutions and notoriously brittle and notoriously low survivability in freeze-thaw cycles so this stuff is not indestructible. They used to market optical media (cd and dvd) as being indestructible, well, let me introduce you to toddlers (and large old toddlers) the optical media for loan at the local public library looks like it was used as drink coasters at a sandpaper factory. Eventually the kids get burned out on unreadable media and stop demanding to borrow it. Anyway, give this "indestructible" media to my local public library and it'll get destroyed in weeks, I bet.

    • (Score: 3, Informative) by turgid on Friday May 09, @09:15PM (2 children)

      by turgid (4318) Subscriber Badge on Friday May 09, @09:15PM (#1403232) Journal

      Maint will never end, you have to keep transferring data every couple years as tech changes.

      Indeed. In the 1980s was the Domesday Project, which involved the BBC Microcomputer and Laser Disc players.

      • (Score: 2, Interesting) by anubi on Saturday May 10, @09:51AM

        by anubi (2828) on Saturday May 10, @09:51AM (#1403282) Journal

        When the first WORM drives ( PRIAM WC-525 ) for the PC (286-AT) came out, I felt I just had to have one to keep some critical design files on.

        It didn't last long. I think it was due to dust collecting on the optics.

        First came the errors and re-read...then it would not work at all. Most of my floppy disks of that era still work. However the only drives that still work for me are of a design that used a specific design of a direct-drive spindle motor and taut-band track positioner.

        I believe TEAC made them.

        --
        "Prove all things; hold fast that which is good." [KJV: I Thessalonians 5:21]
      • (Score: 2) by aafcac on Monday May 12, @04:30PM

        by aafcac (17646) on Monday May 12, @04:30PM (#1403572)

        TBH, that's part of what makes me hesitant about using some of these newer long lived media formats. There's no guarantee that the hardware will be useful down the road and they're often times far less efficient in terms of verifying that the media is still good. It can be a significant pain to just go through and swap disks to make sure they're still working as intended.

    • (Score: 5, Interesting) by jb on Saturday May 10, @05:59AM (1 child)

      by jb (338) on Saturday May 10, @05:59AM (#1403271)

      There's still companies where the fastest bandwidth for data transfer is shipping media.

      "Never underestimate the bandwidth of a station wagon full of mag tape."

      The saying was coined in the days when computing bureaux still existed, but still holds true today.

      • (Score: 2) by VLM on Saturday May 10, @03:17PM

        by VLM (445) Subscriber Badge on Saturday May 10, @03:17PM (#1403303)

        IBM LTO 7 tapes come 20 to a plastic blister pack for somewhat over a kilobuck. Its a nifty sealed clear blister pack that sits in a cardboard box, water sealed again, then in another cardboard box. Why doesn't IBM ship 16 or 32 or some other power of 2, or ship 24 which is two dozen? Who knows big blue works in mysterious ways. They come in shipping crates of 20 IIRC.

        IIRC LTO 7 is 6 TB uncompressed (at least the standard IBM media for that drive and era) so thats 120 TB shippable "very near" anywhere on the planet in less than a day if you have enough money for air courier service, etc.

        120 TB in less than 24 hours is a good deal more than 10 gigabits/sec if my mental math is correct. Probably not more than 30 gigabits/sec unless I really messed up the math

        Still pretty legit, very few companies that aren't telcos have more than ten gigs of internet bandwidth that they can tie up for 24 hours.

    • (Score: 2) by DannyB on Monday May 12, @02:23PM

      by DannyB (5839) Subscriber Badge on Monday May 12, @02:23PM (#1403556) Journal

      Maint will never end, you have to keep transferring data every couple years as tech changes.

      You might not have to do it for as long as you were expecting. Maint WILL end, and you won't have to keep transferring data.

      Universe expected to decay in 10⁷⁸ years, much sooner than previously thought [phys.org]

      --
      The only way to stop a bad guy with a can opener is a good guy with a can opener.
  • (Score: 3, Insightful) by Anonymous Coward on Friday May 09, @06:37PM (3 children)

    by Anonymous Coward on Friday May 09, @06:37PM (#1403213)

    Cerabyte claims the result is a medium as durable as ancient hieroglyphs, with a projected lifespan of 5,000 years or more.

    Having a storage medium that lasts 5000 years is completely, 100% useless unless you also do something to ensure all the equipment needed to read it is, after 5000 years, still actually available and (most importantly) working.

    This is a stupidly hard problem to solve even on timescales multiple orders of magnitude shorter. I could give you a 50 year old disk pack in perfect condition and you simply won't be able to do anything to read it. If you are very lucky, there might be a handful of people in the entire world who would (a) know what to do with it, and (b) have working equipment available to read it.

    For a real-life example, in the early 1990s the BBC migrated their entire analogue video archive to D-3 tape, and now has the better part of half a million D-3 tapes in their archive. To date, less than half of this archive has been converted to a newer format. The tape heads in D-3 readers have a finite operational lifetime, and it is believed that there are not enough remaining heads in the entire world to read the entire remaining archive.

    That probably means anything requiring technology more sophisticated than an optical microscope and a mk.1 eyeball to read it is a non-starter. Probably anything that actually has a chance of being useful after 5000 years will basically look like microfiche encased in some protective glass or whatever.

    Then you also have to deal with the problem that probably nobody alive in 5000 years will be able to understand any of today's written languages, so you also need to do something to ensure that what you record has a chance of being understood. Probably at the bare minimum this means duplicating all the information into as many extant languages and scripts as possible.

    • (Score: 5, Interesting) by VLM on Friday May 09, @07:34PM

      by VLM (445) Subscriber Badge on Friday May 09, @07:34PM (#1403219)

      Having a storage medium that lasts 5000 years

      I'll bet you that's a 5000 year MTBF.

      $7K will get you a LTO-rated tape safe that is fireproof for 1.5 hours and holds exactly 70 LTO tapes. They all look about the same and cost about the same amount because they probably all come from the same Chinese factory. They are designed to be double stacked and need somewhat less than two linear feet of wall, something to do with offgassing during a fire event. So in an environment like the one I worked in, you can fit 4 in an aisle, every two feet, storing 280 tapes. Now that room was about 50 feet square, so lets say 25 safes side by side, thats 7000 tapes per aisle. Lets be realistic about the width, two back to back safes would be 4 feet and you need at least two foot aisles to open the door fully, but like I said lets be reasonable, they push utility carts up and down the aisle and often enough have multiple workers possibly in the same aisle, each aisle "has to" take up at least 8 feet, so the 50 foot square room in the DC will have about 6 aisles. At 7000 tapes per aisle, a fully loaded tape room would have 42000 LTO tapes. LTO-8 currently holds 12 TB (yeah I know LTO-9 is "released") so thats 504000 TB uncompressed or half of one exabyte being stored in this room. Finally, something large enough to hold my entire pr0n collection. Also consider that 42K tapes at fifty bucks each in bulk would be like "meh two million bucks" plus the fireproof safes plus the special media room HVAC and security. So it's a very low 8-figure capital budget, which is quite reasonable at a big mainframe datacenter type facility. I bet that total cost from generic office space to working half exabyte storage room would be about $10M and maybe $4M/year to operate (24x7x365 salaries, replacement drives, etc). This is cheaper than AWS which makes sense you can't expect to save money by providing a middleman with huge profits. AWS seems like it would want $24M/yr, I think, for half a exabyte of storage?

      I'll guess this media can be stored even denser than LTO, at vaporware-date-in-the-future. Or maybe not, but I thought a worked out example for LTO was comparable apples to ... genetically engineered apple-substitute flavoring or whatever analogy.

      Anyway, my point is that a medium size, but not famous or noteworthy, financial institution might very realistically, very reasonably, store around 50K pieces of physical media, so a 5K MTBF means about one tape dies per month. Will they detect the failure? Well, probably not immediately but you might be surprised, this place I worked at liked to scrub to verify. Note that some of these tapes can take hours to read, so you need an absolutely thundering number of drives. It might be quite reasonable to have a drive on top of each safe, on average.

      That's a high enough failure rate that they're going to have a dude on staff who specializes in recovery and repair and replacement and have written procedures, but its not going to be a big enough problem to have a full time job title just to handle broken storage media. At a "pretty large datacenter" it'll be big enough to be some dude's second job, every couple weeks he gets to try to recover data and mail back the broken media under warrantee to get a replacement etc

      Note that some things never change. "pro grade tapes" usually have a lifetime warranty and they're expensive enough that people only toss broken ones in the trash in Hollywood movies or at very small time operations. Everywhere else you ship back to Quantum or IBM or whatever and get a "free" replacement. They never guarantee the data and are never liable for damage to the tape drive, which is annoying (and expensive). AFAIK this policy goes back to the 9-track tape days and also used to apply to pro-grade video tape (the weird stuff pros use not consumer-grade camcorders).

    • (Score: 3, Informative) by HiThere on Friday May 09, @09:37PM (1 child)

      by HiThere (866) on Friday May 09, @09:37PM (#1403235) Journal

      Well the real problem is "how much does a recorder/player cost, and how dense is the storage". Once the media can last a few decades, other factors dominate, except for really specialized uses.

      --
      Javascript is what you use to allow unknown third parties to run software you have no idea about on your computer.
      • (Score: 0) by Anonymous Coward on Saturday May 10, @12:30AM

        by Anonymous Coward on Saturday May 10, @12:30AM (#1403250)
        Also what's the speed. e.g. how long does it take to write and read 1TB. For example if it's only 10MB/sec it'll take more than 1 day to write 1TB.
  • (Score: 3, Interesting) by ElizabethGreene on Saturday May 10, @11:13AM

    by ElizabethGreene (6748) on Saturday May 10, @11:13AM (#1403285) Journal

    We're less than 100 years from the invention of the transistor. Given the relative infancy of tech, the idea of a thousand-year electronic data storage medium is kind of silly.

(1)