20TB Hard Drives May Be Made of Glass

Megalith

24-bit/48kHz
Staff member
Joined
Aug 20, 2006
Messages
13,000
As Seagate and Western Digital push towards 20TB, glass substrates could be a critical component of that effort: Hoya Corp. currently supplies glass substrates for 2.5-inch HDDs but believes that the industry will adopt this same material for 3.5-inch high-capacity drives. Glass substrates provide a number of advantages over aluminum, such as rigidity (thinner, lighter platters that can be spun with less energy and packed more tightly together) and less expansion under heat.

One of the advantages of glass substrates is that they have a higher rigidity than aluminum substrates and can be thinner than aluminum substrates, making it possible to increase the number of disks that can be stored in a case and increase the capacity. Hoya has already prototyped 0.5mm- and 0.381mm-thick substrates, enabling to house 10 and 12 substrates, respectively, in a 3.5-inch HDD whose thickness is about 1 inch.
 
Out of curiosity, what would be the data degradation lifecyle on a medium like that?
 
Didn't IBM try this shit before? And the platters would turn to dust if there was any outside vibration present.
yes, yes they did IBM DeskStar series. It led to them selling their entire storage business to Hitachi. I'm pretty sure it was the guys at HardOCP that gave them the nickname "deathstar" drives because once the heads crash (and due to a math error at ibm, they were guaranteed to crash after ~1 year of use) they scrape all of the metallic compound off of the platters, leaving you with a piece of glass not even DriveSavers can do anything with.

http://www.astro.ufl.edu/~ken/crash/
 
yes, yes they did IBM DeskStar series. It led to them selling their entire storage business to Hitachi. I'm pretty sure it was the guys at HardOCP that gave them the nickname "deathstar" drives because once the heads crash (and due to a math error at ibm, they were guaranteed to crash after ~1 year of use) they scrape all of the metallic compound off of the platters, leaving you with a piece of glass not even DriveSavers can do anything with.

http://www.astro.ufl.edu/~ken/crash/

To hit that level of damage, the user would have had to operate that drive for hundreds of hours after the initial failure.

Glass platters are no more susceptible to head crash damage than any other material. Either way the physical impact of the heads will damage the disc surfaces and/or the heads themselves.
 
To hit that level of damage, the user would have had to operate that drive for hundreds of hours after the initial failure.

Glass platters are no more susceptible to head crash damage than any other material. Either way the physical impact of the heads will damage the disc surfaces and/or the heads themselves.

That's funny. I see you never used one of them. In theory you might be right, but in practice, no. These drives would fail very quickly. Sometimes even shortly after they were purchased. (like less than "hundreds of hours" into total use) Or, as mentioned within a year, which would be on the healthy side for these drives. I had several, because at the time, the performance was good, the prices were decent, and they seemed like a good way to go. All of them were dead shortly after though. I can still remember the sounds that they made. It would start with an almost "beeping" sound, followed shortly by that same beep and some scratchy noises. Shortly after your drive would be useless, and your data gone. I worked for companies that had these, I had friends that had them. They were bad. I'm not even saying it was the substrate's fault. I'm just arguing that they could, and often would fail nearly instantly, with no previous signs of failure.
 
That's funny. I see you never used one of them. In theory you might be right, but in practice, no. These drives would fail very quickly. Sometimes even shortly after they were purchased. (like less than "hundreds of hours" into total use) Or, as mentioned within a year, which would be on the healthy side for these drives. I had several, because at the time, the performance was good, the prices were decent, and they seemed like a good way to go. All of them were dead shortly after though. I can still remember the sounds that they made. It would start with an almost "beeping" sound, followed shortly by that same beep and some scratchy noises. Shortly after your drive would be useless, and your data gone. I worked for companies that had these, I had friends that had them. They were bad. I'm not even saying it was the substrate's fault. I'm just arguing that they could, and often would fail nearly instantly, with no previous signs of failure.

The pictures posted in that link though showed ALL the magnetic material worn off the platter surface. That would take a LONG time, considering the actual size of the drive heads alone, and the contact forces involved. You don't scrub a disc clean with just a head crash. It takes a prolonged head crash and continued attempts at operation for hundreds of hours.

That's not the fault of the platter material though. Lots of laptop drives have used glass platters and never had problems.

That particular series of drives mentioned, the DeskStar, did have a ton of issues and were terrible, terrible drives, just like you said. But those were controller and manufacturing issues, not material problems with glass platters.
 
I'm sure those are probably extreme examples. I'm just talking about how fast the drive becomes useless. I don't doubt that it would take some time to scrape them to hell and back.
 
I'm sure those are probably extreme examples. I'm just talking about how fast the drive becomes useless. I don't doubt that it would take some time to scrape them to hell and back.

Oh, I agree. A single quick head crash can take a drive out. (regardless of substrate material) If you get lucky and just damage the platter surface, then you get a ton of bad sectors, but it might be salvageable. If you happen to damage the head itself, your drive is toast. That can happen in a millisecond.
 
Who knows. I'm pretty sure our minds will be transferrable to some less-volatile medium in the future, though.
 
I loved the old glass platters.
When someone would ask me how I made sure their hdd data was unrecoverable all I had to do was bash the drive with a ball peen hammer and let them shake it. ( Or take the cover off and show them the dust.) ;)
 
I loved the old glass platters.
When someone would ask me how I made sure their hdd data was unrecoverable all I had to do was bash the drive with a ball peen hammer and let them shake it. ( Or take the cover off and show them the dust.) ;)

When I was a repair tech at a loacl shop, and a customer wanted to destroy data, we would just drive a stud anchor (really big nail) through the drive with a couple taps from a 5LB sledge, right through the controller. They got the drive back in a ziploc. Glass? Aluminum? Don't matter, job done.

Kyle's .50 Cal method works too. :)
 
20tb. I'm still very afraid to keep so much on one drive. I don't want to lose everything at once.
And even with a backup drive, what a pain.
 
The pictures posted in that link though showed ALL the magnetic material worn off the platter surface. That would take a LONG time, considering the actual size of the drive heads alone, and the contact forces involved.

Probably would not take as long as you might think. Once the magnetic coating begins to flake off it creates a very fine dust (as seen in the photos). Once the internal filter becomes overwhelmed the dust will begin to circulate, contaminating the other heads and platters, acting like sandpaper. As the other platters begin to break down it generates more and more dust and the rate of breakdown increases rapidly. In that particular case there was probably so much dust generated that it eventually bridged and shorted the head logic or physically jammed the head bearings, hence the half scrubbed platters.
 
Probably would not take as long as you might think. Once the magnetic coating begins to flake off it creates a very fine dust (as seen in the photos). Once the internal filter becomes overwhelmed the dust will begin to circulate, contaminating the other heads and platters, acting like sandpaper. As the other platters begin to break down it generates more and more dust and the rate of breakdown increases rapidly. In that particular case there was probably so much dust generated that it eventually bridged and shorted the head logic or physically jammed the head bearings, hence the half scrubbed platters.

Yeah, you'd get some nice sandblasting effect once you started to grind. It'd still take quite a while to clean ALL the platters as clean as shown in those pictures though. I'd imagine those drives were toast within 5 seconds of the initial head crash, and for some reason they keep them spinning for days and days afterwards.

Sounds like a fun science experiment to determine how long it takes to completely clean the platters.
 
Is anyone still buying spindles?
spinners aren't going anywhere soon
I recommend SSD's for all new builds and hd replacements in laptops.
People freak out when their old and slow laptop boots to a usable desktop in 30 seconds.

The NVME M.2 systems I have made are stoopid fast. If you got the money honey, get one.
 
Doesnt anyone remember the ibm deathstar (deskstar) drives from about 12 years ago having glass platters.

It was such a steaming pile of crap ibm was sued and that forced them to get out of the consumer drive space
 
Doesnt anyone remember the ibm deathstar (deskstar) drives from about 12 years ago having glass platters.

It was such a steaming pile of crap ibm was sued and that forced them to get out of the consumer drive space

Yeah, that was discussed earlier...but that was a problem with the controllers and other stuff in those IBM drives, not the material of the platters. Lots of laptop drives have glass platters, and they work fine.
 
Yeah, you'd get some nice sandblasting effect once you started to grind. It'd still take quite a while to clean ALL the platters as clean as shown in those pictures though. I'd imagine those drives were toast within 5 seconds of the initial head crash, and for some reason they keep them spinning for days and days afterwards.

Sounds like a fun science experiment to determine how long it takes to completely clean the platters.
I was thinking the same thing! Put a plexi top on the drive and add a grain of sand before closing it up. Problem with that is that the plexi would probably get covered with the dust and you wouldn't be able to see jack. :(
 
I was thinking the same thing! Put a plexi top on the drive and add a grain of sand before closing it up. Problem with that is that the plexi would probably get covered with the dust and you wouldn't be able to see jack. :(

Setup 5-6 old drives, brand new. Spin them up and force head crashes on all of them at once. Check out one right away, then check out the next one like 4-5 hours later. Then crack another open a day later. Based on that you should get some idea of how long to wait to open the next drives. You'd need a lot more drives to really be scientific about it though.
 
What may be of surprise to most of you, spindle drives

1: Do maintain a cost/MB advantage
2: Are approaching the speed of solid state drives on SATA III (Sequential Access)

That said long live 3D MLC SSD! (Death to TLC and QLC)
 
Is anyone still buying spindles?

Not me. I'm all solid state except one drive in a third machine. I've just learned to keep my usage down. Any old-but-keep data I have is spread across a few large detached Flash drives.
 
Laptops have been using glass as mentioned for over a decade. They're pretty reliable too as long as you're not smashing them around.

I for one welcome our glass overlords. If they do bring 20Tb out, it will do well to replace my 8TB arrays.
 
Is anyone still buying spindles?

In a ssd market that has gone back up in the past year? Long term storage? Nah. Spining rust is dead! Totes man! I mean no one has bandwidth caps, just re download it all bro.
 
Every build I make for people has both. Large Spindle for mass storage or slow garbage you don't care about so much. SSD for the rest or NVME etc.
 
Laptops have been using glass as mentioned for over a decade. They're pretty reliable too as long as you're not smashing them around.

I for one welcome our glass overlords. If they do bring 20Tb out, it will do well to replace my 8TB arrays.

Most hard drives reliability drops rapidly when you smash them around, regardless of material.

It's worth it to mention they're not pure glass...they're glass and ceramic mixed, and they've got shock ratings in the same category as any other hard drive.
 
glass platter hard drives have been around since at least the mid 90s typically used in laptop drives. glass is very stable when it comes to thermal/rotational loads. Not to mention it is lighter and easier to spin up
 
I am special in my company as I have the only PC with a SSD and our network performance still makes this machine feel ultra slow....
 
Back
Top