Google asked them to make drives with more platters and larger platter sizes ages ago... We could have had 100TB drives by now. An extra 0.5 inches makes a large difference in area. Nobody cares about maintaining compatibility with 3.5" drives which are ancient. Also, what's with consumer pricing never falling below around $50/2TB, the price back in 2009?
changing the physical format of the drives would be a huge issue as everything is setup for that form factor. From drive enclosures to the servers themselves, even half an inch/~12mm, would cause chaos in the ecosystem and would take years to fully take effect. Remember, even the new EDSFF SSD formats (E1/E3) are taking a while to get incorporated into the drive pools because servers needed redesigns.
3.5" drive spec allows for up to 41mm height. Not that anyone has made that in some 2~3 decades. We're talking back when Maxtor was still in business and there wasn't an ATX standard.
Like I said in my first comment: nobody really cares about the standardization, it is an artificial limitation. That just means they decided not to do it yet. You can make a new standard if you so choose. It doesn't need to "fully take effect", it just needs to exist. Datacenter customers (like Google, who suggested the changes) would appreciate the 50% reduction in drive count, controllers, etc. A diameter increase of 0.5" would increase drive sizes by 34%, if you add more platters, 50% or more.
The edge of a 7200rpm 3.5" disk is already moving at around 120km/h, and the drive heads have to float on this disk at 5nm height.
If you increase the platter diameter, it's going to change quite a lot of other factors on the disk. For instance a longer actuator arm, stronger motor. Maybe even stronger platters. I just don't think it's as simple as increasing disk diameter.
what happens with the price…simply no competition. idiotic antitrust authorities around the world allowed the market to consolidate in 3 competitors that simply agree prices between themselves
Yes, some people, but not all. By adding more platters, I'm talking about stretching the drive height-wise. And that 100TB would become 150-200TB with a larger drive. So far, I haven't heard any real reasons not to make each drive larger. There's a reason we don't use servers full of 1.8" iPod HDDs.
50TB in 2 years from a company that was 3 years late delivering 22TB just last year. So they think they can more than double capacity in 2 years?
Companies are so ridiculous now. Their product disclosures and timelines are outright lies, containing technology they say is on schedule that is destined to be vaporware (or like HAMR, a decade late.) It's as if nobody cares about actual product execution anymore. Look at Tesla. 5 years late on Cybertruck. That's an entire lifecycle for a vehicle. Cybertruck should be on V2 by the time V1 materializes.
If you want the highest capacity (for enormous cost) SSD (at 100TB, or 50 TB SSDs for a significant savings) beats an HDD.
Now you can say just buy 2 or 3 HDDs and you're at the same capacity for a much lower cost, but that only considers cost, and nothing else (except capacity).
To make it more even, and using a useful measurement, you need capacity and access speed, because in Enterprise the drives are likely to be in RAID to increase access speed and protect the data.
I calculate that as asking do you want 15 HDDs (RAIDed, somehow) or two SSDs for the same cost, and access speed (when in a RAID array that would increase the HDD's speed 15x or the SSD's capacity 2x); resulting in equal capacity and access speed for the same cost.
Seems like SSDs win, except for endurance; which is traded for space and electricity.
HDD cost must drop, quicker than SSD will comparatively, for the race to be equal. I'm betting I know who'll win; unless HDDs pull off PB capacity, as I don't see the access speeds quadrupling by any means.
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
Facebook
Twitter
RSS
11 Comments
Back to Article
flyingpants265 - Wednesday, November 8, 2023 - link
Google asked them to make drives with more platters and larger platter sizes ages ago... We could have had 100TB drives by now. An extra 0.5 inches makes a large difference in area. Nobody cares about maintaining compatibility with 3.5" drives which are ancient.Also, what's with consumer pricing never falling below around $50/2TB, the price back in 2009?
HaninAT - Wednesday, November 8, 2023 - link
changing the physical format of the drives would be a huge issue as everything is setup for that form factor. From drive enclosures to the servers themselves, even half an inch/~12mm, would cause chaos in the ecosystem and would take years to fully take effect. Remember, even the new EDSFF SSD formats (E1/E3) are taking a while to get incorporated into the drive pools because servers needed redesigns.meacupla - Wednesday, November 8, 2023 - link
3.5" drive spec allows for up to 41mm height.Not that anyone has made that in some 2~3 decades.
We're talking back when Maxtor was still in business and there wasn't an ATX standard.
flyingpants265 - Friday, November 10, 2023 - link
Like I said in my first comment: nobody really cares about the standardization, it is an artificial limitation. That just means they decided not to do it yet. You can make a new standard if you so choose. It doesn't need to "fully take effect", it just needs to exist. Datacenter customers (like Google, who suggested the changes) would appreciate the 50% reduction in drive count, controllers, etc.A diameter increase of 0.5" would increase drive sizes by 34%, if you add more platters, 50% or more.
meacupla - Friday, November 10, 2023 - link
The edge of a 7200rpm 3.5" disk is already moving at around 120km/h, and the drive heads have to float on this disk at 5nm height.If you increase the platter diameter, it's going to change quite a lot of other factors on the disk.
For instance a longer actuator arm, stronger motor. Maybe even stronger platters.
I just don't think it's as simple as increasing disk diameter.
lucaB75 - Wednesday, November 8, 2023 - link
what happens with the price…simply no competition. idiotic antitrust authorities around the world allowed the market to consolidate in 3 competitors that simply agree prices between themselvesJames5mith - Wednesday, November 8, 2023 - link
Consumers aren't driving larger drives, enterprise is. And enterprise customers can eat the higher costs. There is no reason for it to become cheaper.nandnandnand - Wednesday, November 8, 2023 - link
Some people would definitely care about compatibility, and more platters have been added to helium-filled drives.We could have 100 TB drives without resorting to making them larger if HAMR hadn't been about a decade behind schedule.
flyingpants265 - Friday, November 10, 2023 - link
Yes, some people, but not all. By adding more platters, I'm talking about stretching the drive height-wise. And that 100TB would become 150-200TB with a larger drive.So far, I haven't heard any real reasons not to make each drive larger. There's a reason we don't use servers full of 1.8" iPod HDDs.
Samus - Saturday, November 11, 2023 - link
50TB in 2 years from a company that was 3 years late delivering 22TB just last year. So they think they can more than double capacity in 2 years?Companies are so ridiculous now. Their product disclosures and timelines are outright lies, containing technology they say is on schedule that is destined to be vaporware (or like HAMR, a decade late.) It's as if nobody cares about actual product execution anymore. Look at Tesla. 5 years late on Cybertruck. That's an entire lifecycle for a vehicle. Cybertruck should be on V2 by the time V1 materializes.
Rοb - Monday, November 13, 2023 - link
If you want the highest capacity (for enormous cost) SSD (at 100TB, or 50 TB SSDs for a significant savings) beats an HDD.Now you can say just buy 2 or 3 HDDs and you're at the same capacity for a much lower cost, but that only considers cost, and nothing else (except capacity).
To make it more even, and using a useful measurement, you need capacity and access speed, because in Enterprise the drives are likely to be in RAID to increase access speed and protect the data.
I calculate that as asking do you want 15 HDDs (RAIDed, somehow) or two SSDs for the same cost, and access speed (when in a RAID array that would increase the HDD's speed 15x or the SSD's capacity 2x); resulting in equal capacity and access speed for the same cost.
Seems like SSDs win, except for endurance; which is traded for space and electricity.
HDD cost must drop, quicker than SSD will comparatively, for the race to be equal. I'm betting I know who'll win; unless HDDs pull off PB capacity, as I don't see the access speeds quadrupling by any means.