I bought computer hardware for Motorola in 1980. The standard disk then was the CDC Storage Module Drive. 300mb and the size of a washing machine, sold for $15,000.
My home PC has 3TB of hard drive, so that's 9,000 washing machines at $15,000 each = $135 MILLION in 1980 dollars (and a 5 story building probably).
edit: my avatar picture is a disk platter from a CDC 808 disk drive. You can see it is about twice as tall as a regular hammer. It held 5mb. The drive had 12 of these to make up a 60mb disk, about the size of a double-wide refrigerator/freezer.
... neither of which really addresses the question of the thread ?
You just repeat the good old "nobody will ever need more than 1 Megabyte of RAM" line of argument that Bill Gates may or may not have uttered in the 1980s.
They do when they need to. Extremely large drives are very niche.
Most data centers won't use 20tb hard drives unless it is for archives or backups. Each drive supports only so many iops. If you build a raid group with 20tb drives you will get half the iops as one with 10tb drives.
In most cases iops are just as important as raw data capacity.
I suspect we will see ssd continue to grow and push way beyond platter drives. Platter drives are too slow so I am not sure it will ever make sense to grow them again. If the need arises it will happen. Ssd have more iop capacity per drive so they will grow beyond the 60tb most likely. I could see 100tb ssd in the 15 years.
“It's unwise to pay too much, but it's worse to pay too little. When you pay too much, you lose a little money - that's all. When you pay too little, you sometimes lose everything, because the thing you bought was incapable of doing the thing it was bought to do. The common law of business balance prohibits paying a little and getting a lot - it can't be done. If you deal with the lowest bidder, it is well to add something for the risk you run, and if you do that you will have enough to pay for something better.”
Data centers hate having to rebuild RAID arrays when a huge hard drive fails. SSDs have gotten massively faster over the last decade. Hard drives have only gotten a little faster. It's one thing to take an hour to rebuild a RAID array when a large SSD fails. It's quite another to take a day to rebuild it when a large hard drive fails.
It's not just the time it takes. It's also that you have diminished redundancy while rebuilding the array, and that means an increased risk that you could lose the data entirely. And data centers really, really, really hate losing their data entirely.
I suspect we will see ssd continue to grow and push way beyond platter drives. Platter drives are too slow so I am not sure it will ever make sense to grow them again. If the need arises it will happen. Ssd have more iop capacity per drive so they will grow beyond the 60tb most likely. I could see 100tb ssd in the 15 years.
I've had my current drive since 2019 and it hasn't grown at all. I put a little mark on the wall to track the little guy's growth but no matter what data I feed him he is still the same size at when we first brought him home.
It's OK, we love him just the same.
All time classic MY NEW FAVORITE POST! (Keep laying those bricks)
"I should point out that no other company has shipped out a beta on a disc before this." - Official Mortal Online Lead Community Moderator
Proudly wearing the Harbinger badge since Dec 23, 2017.
Coined the phrase "Role-Playing a Development Team" January 2018
"Oddly Slap is the main reason I stay in these forums." - Mystichaze April 9th 2018
... neither of which really addresses the question of the thread ?
You just repeat the good old "nobody will ever need more than 1 Megabyte of RAM" line of argument that Bill Gates may or may not have uttered in the 1980s.
But you do you, I guess.
It sort of does address the point in a roundabout manner. There is very limited demand for very large spinning platters. Disk drives don't scale to size as well (easily) as nand drives do and the cost and failure rate goes up. Is it technically possible? Probably (most certainly), but it doesn't make good business sense to build a product not fit for most consumer use.
There is a lot of good information on size and failure rate which is important even for consumers buying large drives, typically for NAS use. The largest drive size they use is 16TB.
You didn't really mention the use case for the drive, which is relevant and important. Assuming you want to drop it in a NAS or desktop, many (most or all?) consumer NAS have a max raw storage limit and a max single volume size. For example, most synology servers aren't going to consume a 20TB drive in their raid array because the max single volume exceeds the 16 or 18GB max for the model. In a consumer desktop there is no practical per volume storage limit if your bios supports GPT partitioning. However, the number of drive slots is limiting.
So, it isn't going to change anytime soon because most people want something with a better $$/space ratio and reliable failure rate. That doesn't even factor in nand getting so much cheaper.
Well, right now SSDs are still about five times more expensive than harddrives, per storage unit. So yeah at some point SSDs will be as cheap as harddrives, and that will be the end of harddrives. It will be as great a development as when the old tubes got replaced by flatscreens. But I would guess its about a decade in the future or some such ?
And why yes, I meant to write TB, not GB. Sorry about that one, clearly wasnt paying attention.
The maximum harddrive size you can get is currently actually 22 TB, but its only one company that offers that. Everyone else stops at 20 TB, maximum. But that was the maximum years ago, already, even before COVID19.
It's not just the time it takes. It's also
that you have diminished redundancy while rebuilding the array, and
that means an increased risk that you could lose the data entirely. And
data centers really, really, really hate losing their data entirely.
I would assume they all use RAID 6 to protect against that.
[...] For
example, most synology servers aren't going to consume a 20TB drive in
their raid array because the max single volume exceeds the 16 or 18GB
max for the model. [...]
The only way that I expect SSDs to ever be as cheap as hard drives on a capacity basis is if hard drives stop making progress on technology but SSDs keep going for quite a while after that.
RAID 6 offers quite a bit of data protection, but it's not invincible. Once you replace a drive and start rebuilding the array, if two more drives die while you're rebuilding the array, your data is gone. Rebuilding the array hammers on the other drives really hard, so that makes it more likely that they fail than if the system were mostly idle. The longer it takes to rebuild the array, the more risk you have of additional failures causing you to lose data.
Yes, RAID 6 usually works and lets you rebuild the array just fine. But some degrees of "usually" aren't good enough for some purposes.
If $40K for a 100 TB SSD is too much for you, here's a really fast 30.72 TB SSD for under $4K:
I bought computer hardware for Motorola in 1980. The standard disk then was the CDC Storage Module Drive. 300mb and the size of a washing machine, sold for $15,000.
My home PC has 3TB of hard drive, so that's 9,000 washing machines at $15,000 each = $135 MILLION in 1980 dollars (and a 5 story building probably).
edit: my avatar picture is a disk platter from a CDC 808 disk drive. You can see it is about twice as tall as a regular hammer. It held 5mb. The drive had 12 of these to make up a 60mb disk, about the size of a double-wide refrigerator/freezer.
For years now I've been wondering what that was; thought it was a record. Finally the the mystery has been answered.
I'll just say it, the investment for hard discs has dried up. There isn't a return on investment to get more bits out of a disc. That investment has moved on to other storage solutions.
The only way that I expect SSDs to ever be as cheap as hard drives on a capacity basis is if hard drives stop making progress on technology but SSDs keep going for quite a while after that.
But thats exactly what I'm pointing out ? We are stuck at 20 TB max since how many years now ? We definitely already had them before COVID19.
I have 10TB and I have 90% still free. I usually use google drive for a lot of things. Even so, I doubt it would approach that much space. To be fair, I'm a chronic organizer, so I try to make sure to remove programs and files that I don't use anymore so that my computer isn't cluttered. I'm the same with my home screen. Mine is probably emptier than 99% of the people I know.
Hell I don't know what you guys need all that storage for... your giant donkey pr0n collection I'm guessing. I have a total of 4TB, 1TB SSD and 3TB old school spinny disk, and while I do actually have to uninstall games I'm not playing to manage the SSD size, I've never had an issue having all the stuff I'm actually using/playing installed at the same time. Maybe go through on occasion and delete all the movies you illegally downloaded from your hacker sites that you aren't currently watching?
Heat-assisted magnetic recording (HAMR) increases the storage capacity of HDDs by utilizing a nanoscopic laser diode attached to the read/write head. The laser diode creates and parses smaller data bits that are still magnetically stable.
Comments
First, unless you are a data center operator, is 20TB not big enough for almost every single consumer user out there?
Second, unless you are a data center operator or a hoarder, aren't most consumer users shifting everything over to SSD storage anyway?
------------
2024: 47 years on the Net.
--John Ruskin
It's not just the time it takes. It's also that you have diminished redundancy while rebuilding the array, and that means an increased risk that you could lose the data entirely. And data centers really, really, really hate losing their data entirely.
https://nimbusdata.com/products/exadrive/specifications/
100 TB over SATA is pretty niche, but more mainstream U.3 drives currently go up to 30.72 TB.
It's OK, we love him just the same.
All time classic MY NEW FAVORITE POST! (Keep laying those bricks)
"I should point out that no other company has shipped out a beta on a disc before this." - Official Mortal Online Lead Community Moderator
Proudly wearing the Harbinger badge since Dec 23, 2017.
Coined the phrase "Role-Playing a Development Team" January 2018
"Oddly Slap is the main reason I stay in these forums." - Mystichaze April 9th 2018
RAID 6 offers quite a bit of data protection, but it's not invincible. Once you replace a drive and start rebuilding the array, if two more drives die while you're rebuilding the array, your data is gone. Rebuilding the array hammers on the other drives really hard, so that makes it more likely that they fail than if the system were mostly idle. The longer it takes to rebuild the array, the more risk you have of additional failures causing you to lose data.
Yes, RAID 6 usually works and lets you rebuild the array just fine. But some degrees of "usually" aren't good enough for some purposes.
If $40K for a 100 TB SSD is too much for you, here's a really fast 30.72 TB SSD for under $4K:
https://www.newegg.com/Micron-30-72-TB-9400/p/N82E16820363150
That's also bigger than 20 TB, and it comes in a 2.5" form factor, too.
For years now I've been wondering what that was; thought it was a record. Finally the the mystery has been answered.
Heat-assisted magnetic recording (HAMR) increases the storage capacity of HDDs by utilizing a nanoscopic laser diode attached to the read/write head. The laser diode creates and parses smaller data bits that are still magnetically stable.