Do you think the hard drive industry will return to half height (1.63" tall) 3.5" hard drives?

[cbn]

The current standard for 3.5" hard drives is 1" tall, but in the past there also was a half height (1.63" tall) standard as well.

http://www.pcguide.com/ref/hdd/op/formIn35-c.html

3.5" form factor drives come in two general profiles: the larger is the so-called half-height drive, which is 1.63" in height. This name is kind of funny, since it is "half" of a height that never existed for 3.5" form factor drives. The name was derived from the fact that these drives are the same height as half-height 5.25" form factor drives, which are half the height of full-height 3.25" high drives in that form factor. Half-height 3.5" form factor drives are still used today, but only in servers and other high-end platforms. The standard for 3.5" is 1" height, which is commonly called slimline or low-profile, but just as commonly given no name at all and assumed as the default. The reason for the smaller size being the standard is that 1" is the height of a standard 3.5" floppy disk drive and 3.5" drive bay. In addition, there are some drives that are reduced in size from the 1" standard, using for example 0.75" height.

I'm thinking with multi-actuators on the way the taller 1.63" drives have a higher chance of making a comeback.

If not 1.63" at least 1.5".

 

[dave_the_nerd]

Nah. All those rackmount enclosures are built around 1" drives. You increase the height of the drives, you cut spindle count dramatically, no net performance increase per deployed U of rackspace. And you have to buy all new enclosures. Gross.

 

[Red Squirrel]

Yep best to stick to a standard 1" than to change it. I could maybe see it happen at some point but hopefully not... Paid a lot of money for my enclosure don't want to change it any time soon.

 

[cbn]

Nah. All those rackmount enclosures are built around 1" drives. You increase the height of the drives, you cut spindle count dramatically, no net performance increase per deployed U of rackspace. And you have to buy all new enclosures. Gross.

1. Are you sure the rackmount enclosures can't be modified? Which ones are you thinking about?

I know the BackBlaze POD could be modified:

http://www.storagereview.com/backblaze_storage_pod_60_review

https://www.backblaze.com/blog/open-source-data-storage-server/ ( As it is an open source design this article includes 2-D Blueprints, 3-D Solidwork files, STEP files, Drive Guide design files, Storage Pod 6.0 Parts list, Wiring Diagrams, Wiring routes, and a really good build book which has some very detailed pictures in it)

2. In how many scenarios are enclosures re-used vs. buying a new storage server?

P.S. 1U server with four 3.5" 1.63" bays could have greater capacity (per U) as there is room at the top for taller drives and I believe there are Workstations and desktops that could accommodate 1.63" (or at least 1.5") tall 3.5" drives.

 

[dave_the_nerd]

1. Are you sure the rackmount enclosures can't be modified? Which ones are you thinking about?

Stuff like this: http://www.istarusa.com/istarusa/products.php?model=E-212L

Or this:

http://www.raidinc.com/products/jbod/4u-84-bay-jbod

I agree having to buy new enclosure is a definite negative. The drives would have to be sufficiently cheaper to offset the cost of the new rack otherwise 1" drive would be used instead.

With that noted, I know the BackBlaze POD could be modified:

http://www.storagereview.com/backblaze_storage_pod_60_review

https://www.backblaze.com/blog/open-source-data-storage-server/ ( As it is an open source design this article includes 2-D Blueprints, 3-D Solidwork files, STEP files, Drive Guide design files, Storage Pod 6.0 Parts list, Wiring Diagrams, Wiring routes, and a really good build book which has some very detailed pictures in it)

Sure, they could adjust the spacing and put fewer disks in it. But they'd still have to rebuild their pods.

Most companies aren't going to "modify" their hardware. That's not how corporate IT works.

2. In how many scenarios are enclosures re-used vs. buying a new storage server?

Whenever possible. As long as you're not upgrading your interconnect fabric. 6Gb SAS is from 2009 and is still plenty common (although 12Gb is the new hotness for new installations.)

P.S. 1U server with four 3.5" 1.63" bays could have greater capacity (per U) as there is room at the top for taller drives and I believe there are Workstations and desktops that could accommodate 1.63" (or at least 1.5") tall 3.5" drives.

1U is only 1.75". A 1.63" HDD wouldn't fit, once you add in a caddy and stuff.

Desktops will get hand-me-down cheapified server/enterprise tech, as usual. Desktop applications have never really driven the market.

 

[cbn]

So maybe 1.5" for SAS 24 Gbps installations?

4 1.5" tall 3.5" drives per 1U server
8 1.5" tall 3.5" drives per 2U server
40 1.5" tall 3.5" drives per 4U Top Loader server

Also 1.5" tall 3.5" drives for select workstations (I think HP Z4, Z6, Z8 could fit this size).

P.S. Extrapolating from this article I reckon Seagate could get at least 18 .381mm glass platters in a 1.5" tall 3.5" hard drive. With Six actuators, 7200rpm and 3TB platters that should saturate a single lane of SAS 24 Gbps. With further increases in areal density and/or more actuators allowing the saturation of wide port (two lanes) of SAS 24 Gbps.

The company has already prototyped glass substrates measuring 0.5mm and 0.381mm thick, which would allow a dozen substrates in a single 3.5-inch HDD

 

[dave_the_nerd]

You're funny.

 

[eton975]

I think the question will be: will there be enough of a density improvement and economic incentive for Seagate to do it and drive farms to buy said drives? I suspect the answer is no for the near future, personally.

 

[cbn]

I think the question will be: will there be enough of a density improvement and economic incentive for Seagate to do it and drive farms to buy said drives? I suspect the answer is no for the near future, personally.

HAMR is expensive (apparently more expensive than MAMR) so I think there is incentive for Seagate to do this. (Larger drives are cheaper per TB to make than smaller drives). In fact, the company appears to be moving in that direction.

https://www.anandtech.com/show/11037/seagate-to-shut-down-one-of-its-largest-hdd-assembly-plants

It is important to note that while Seagate cuts down HDD assembly capacities, it does nothing to plants that produce heads, media and substrates. Over the past few years, unit shipments of HDDs have declined, but their average capacities increased (especially capacities of HDDs for datacenters) due to strong demand for high-capacity SKUs. Therefore Seagate may not need to produce a lot of drives, but it needs to pack about the same amount of heads and platters into fewer HDDs. Moreover, in the coming years, the company will need more heads because of new manufacturing technologies (TDMR, HAMR, etc.) and more media because it can now pack more platters into high-end helium-filled drives.

You're funny.

Regarding upgrading interconnect fabric (which you state as reason to change enclosures) I think there is still some time for SAS 24 Gbps as the first end user products don't even launch for at least another year and a half. But if doesn't happen by that time there is always Next Gen SAS (which I assume to be SAS 48 Gbps based on PCIe 5.0) to look forward to.

P.S. SAS Roadmap from this Dec 5th 2017 webinar.