What storage form factor due you think will decrease in usage the most over the next 5 years?

[Red Squirrel]

Spindle drives are still viable for mass storage so they arn't going away until SSDs can become same cost per TB and not wear out. That said I could see 3.5 slowly get phased out as higher capacity spindle drives are made in 2.5 factor. The other ones arn't really viable for mass storage as you are greatly limited by slots but they will probably stay around as they are great for a boot drive especially when building a small form factor. One less piece of hardware that has to be attached somewhere and one less cable. (well two, I guess)

The idea of a mass storage array using 2.5 drives is kinda sexy. They can be all vertical and you can fit like 24+ in a 3U chassis.

 

[dave_the_nerd]

Spindle drives are still viable for mass storage so they arn't going away until SSDs can become same cost per TB and not wear out.

Even accounting for "wearing out" SSDs are more reliable than spinners. Stop repeating that 2009-era FUD.

That said I could see 3.5 slowly get phased out as higher capacity spindle drives are made in 2.5 factor. The other ones arn't really viable for mass storage as you are greatly limited by slots but they will probably stay around as they are great for a boot drive especially when building a small form factor. One less piece of hardware that has to be attached somewhere and one less cable. (well two, I guess)

The idea of a mass storage array using 2.5 drives is kinda sexy. They can be all vertical and you can fit like 24+ in a 3U chassis.

2U, actually.

In terms of TB/U the big 5U 84-bay 3.5" drive enclosures still beat it, but the perf/watt is kinda face-peeling.

 

[cbn]

Even accounting for "wearing out" SSDs are more reliable than spinners.

According to this study (using drives in Google's data center) the replacement rate is lower for SSD, but SSD is more likely to lose data.

https://www.zdnet.com/article/ssd-reliability-in-the-real-world-googles-experience/

The SSD is less likely to fail during its normal life, but more likely to lose data.

 

[cbn]

That said I could see 3.5 slowly get phased out as higher capacity spindle drives are made in 2.5 factor. The other ones arn't really viable for mass storage as you are greatly limited by slots but they will probably stay around as they are great for a boot drive especially when building a small form factor. One less piece of hardware that has to be attached somewhere and one less cable. (well two, I guess)

The idea of a mass storage array using 2.5 drives is kinda sexy. They can be all vertical and you can fit like 24+ in a 3U chassis.

2U, actually.

In terms of TB/U the big 5U 84-bay 3.5" drive enclosures still beat it, but the perf/watt is kinda face-peeling.

Looking at the Seagate Roadmap below it looks around 2022 to 2023 the multi-actuator 40+ TB drives will be using lots of actuators (12 per drive/1 per glass platter) and very low RPM if the scaling in the graph holds up being true:

That would really increase performance per watt of 3.5" a lot!

With that mentioned, nothing is to prevent Seagate from using multi-actuators and low RPM on 2.5" as well. However, 3.5" will likely be much more economical.....and actually they could even go dual pillar on 3.5" which I reckon would use ~3" platters and up to 24 actuators for even lower RPM, but more IOPS and Sequential.

 

[Red Squirrel]

Wow that's really neat. I always wondered if multi actuator would ever be a thing. The thing with mechanical drives, because they have moving parts they have a CHANCE to fail at any given time, but the actual medium does not wear out over time. So in theory a spindle drive will last the same amount of time no matter how much IO is thrown at it. (of course it could die from heat or other cause that's not directly related to the IO itself). The wear on SSDs is not FUD, it's just the nature of how flash works. Unless they've made some huge advancements that I'm not aware of. If I look at the smart status of my OS SSD it's at 3% usage. That's from casual OS usage mind you, no mass data IO like VMs or databases or raid etc. So for the purpose of an OS drive it's probably going to outlive it's useful life as I'm bound to upgrade it at some point before it dies. I would be curious to try a raid array with some SSDs though, they are coming down in price, you can get 1-2TB ones for under a grand now. I'd be reluctant to do raid 10 because of cost but probably do raid 5.

 

[BonzaiDuck]

I've seen laptop-class 2.5" HDDs up to 4TB capacity (a Seagate Barracuda). But look at all the factors that play into this.

Excluding a reference to server farms, 3.5" still represents a faster (though limited) speed potential and storage size beyond 10TB. If you have a need for a 10TB+ HDD, you'll still buy a 3.5". Three, four, five years from now -- how will that change? Assume the capacity maximum for 2.5" will slowly increase.

How long will it be before we see 4 and 6TB SATA SSDs? How long before NVME M.2 will show up with the same high capacity?

The reason I migrated from 3.5" drives to 2.5" HDDs in my desktop/workstation system was weight and power consumption. You take a hit on speed, dropping to 5,400 rpm from 7,200. But with 3D Xpoint, more RAM and dedicated NVME caching drives, none of that matters. You only would care about capacity and dollars per GB. That assumes your boot drive is either SATA or NVME SSD. I wouldn't build a system around an HDD boot-disk. Who would?

 

[IntelUser2000]

Even accounting for "wearing out" SSDs are more reliable than spinners. Stop repeating that 2009-era FUD.

SSDs, when it fails are much worse because it doesn't happen gradually, but suddenly. And the above article with more bit error rates are worrying too. Failed HDDs can easily recover data. SSDs fails completely, so often the system doesn't even recognize it.

Of course it shouldn't worry most people.

 

[rsutoratosu]

SSDs, when it fails are much worse because it doesn't happen gradually, but suddenly. And the above article with more bit error rates are worrying too. Failed HDDs can easily recover data. SSDs fails completely, so often the system doesn't even recognize it.

Of course it shouldn't worry most people.

which means backup asap ! I had a nvme drive from dell that gave me bad sectors last night.. didn't get a chance to look at it but I was wiping it for windows reinstall, this on a laptop I rarely used. Drive is 2 years old but I didn't get a chance to run crystal disk on it to see what the stats are

 

[dave_the_nerd]

SSDs, when it fails are much worse because it doesn't happen gradually, but suddenly. And the above article with more bit error rates are worrying too. Failed HDDs can easily recover data. SSDs fails completely, so often the system doesn't even recognize it.

Of course it shouldn't worry most people.

People who worry about how many 2.5" drives you can fit in a 2U enclosure aren't going to run data recovery software, or waste time copying data off of a failing drive. We toss the drive in a shredder when SMART says to, and rebuild the array or (worst-case) restore from a backup. It's much faster than fiddling around with some recovery tool.

 

[IntelUser2000]

People who worry about how many 2.5" drives you can fit in a 2U enclosure aren't going to run data recovery software, or waste time copying data off of a failing drive. We toss the drive in a shredder when SMART says to, and rebuild the array or (worst-case) restore from a backup. It's much faster than fiddling around with some recovery tool.

Going back to the original post, and tying in your previous comment, you can find counterpoints to nearly every discussion and argument, but talking about the whole market, the highest volume is very likely in single, or few drive systems.

Outside of servers, very little consumers will go for many SSDs as storage. They'll be using HDDs anyway. Making the point kinda moot.

I don't think HDDs will ever become a tiny portion of the market because improvements in magnetic recording is proving to be just as scalable as vertical stacks and process tech improvements for SSD storage.

 

[mopardude87]

I casted a vote for the 3.5. Eventually the solid state will become so much more cheaper its 2.5 will make for easier fits and smaller overall builds. Heck for the last decade i never used over 500gb till recently cause every other game is the size of the RMS Titanic all of a sudden really 80gb for BF1 without any dlc?. I could alone can get by on a 1tb ssd install all my games and then be done with any sort of hard drive upgrade for the foreseeable future. The games i play i get bored of get removed and one new one goes back in its place.

 

[fleshconsumed]

Depends on the market. 2.5 is a very convenient format for servers because you can fit a lot of them in a 2U server, so I can see it staying there for a long time. However, when it comes to consumer market 2.5 is quickly becoming useless. 2.5 original purpose was laptops, now that almost every laptop is moving to M.2 and almost every desktop motherboard comes with M.2 slot there is no point to 2.5 aside from cost. Once M.2 price premium disappears, so will the 2.5 drives. 3.5" will always have a role as massive cheap storage that no other format can approach so it'll have staying power too.

So overall I think 2.5 will decline first (although it will linger for quite a while longer in server market until flash prices come down), followed by 3.5 since most consumers do not need massive storage, and I expect M.2 to become primary storage format until something better comes along.

 

[HutchinsonJC]

However, when it comes to consumer market 2.5 is quickly becoming useless.

You must've missed someone's comment about them being pretty popular for backup external drives. I mean, I rotate through 10 different 2.5" form factor drives for backups, as an example.

 

[cbn]

How long will it be before we see 4 and 6TB SATA SSDs? How long before NVME M.2 will show up with the same high capacity?

4TB 2.5" SATA SSD has been made by Samsung since 2016:

https://www.anandtech.com/show/10481/the-samsung-850-evo-4tb-ssd-review

So 4TB M.2 NVMe is probably not that far away,

However, with that mentioned it does appear that 2.5" 7mm SSD has 2x the capacity as M.2 2280 all things being equal.

 

[cbn]

Looking at the lack of NVMe scaling for consumer workloads mentioned in this post and this post (with this post showing an exception) I got to wondering about the future of PCIe 4.0 x 4 or 5.0 x 4 NVMe SSDs.

Could it be that PCIe 4.0 x 2 or even PCIe 5.0 x 1 (or PCIe 5.0 x2) becomes more popular that expected?

If so, then maybe the 2.5" form factor gets even more lifespan than predicted due the versatility of its backplane.

Looking at the U.2 (SFF-8639) data and power connector:

It can handle PCIe x 4 or PCIe x2 + SATA x 2. This, in addition to SAS.

So a laptop could have U.2 7mm, SATA Express or SATA drive could all plugged into the same 2.5" bay.

P.S. Something else to think about for the future of SATA Express (and SATA) would be SATA 12 Gbps using surplus PCIe 4.0 or PCIe 5.0 lanes leftover from the four PCIe 4.0 or 5.0 lanes. (Remember SATA Express only uses 2 PCIe lanes for the solid state portion of the drive).

 

[cbn]

In reference to what I wrote in post #29, below I have linked a drive called Seagate Exos 7E2000:

https://www.seagate.com/enterprise-storage/exos-drives/exos-e-drives/#exos-7e2000

It is 2.5" x 15mm 7200 rpm drive with a capacity of either 1TB or 2TB. This for Warm storage.

I believe these are populated in a server rack 24 per 2U.

I am assuming a multi-actuator version of these (4TB with 2 actuators, 8TB with 4 actuators, etc) has or is being considered. This to increase data density.

So the question is 3.5" dual pillar multi-actuator vs. 2.5" single pillar multi-actuator.

I think they are both very interesting. Dual pillar 3.5" (at the same number of platters as single pillar 3.5") sacrifices some data density (when used in a 12 3.5" bay 2U enclosure) due to the smaller ~3" platter reducing storage capacity. However, as mentioned by dave_the_ nerd, there are 84 bay 3.5" in 5U (or in the case of Back Blaze Pod 60 bay 3.5" in 4U) enclosures available. This regains the lost data density.

Now regarding IOPs consider that because the 3.5" has dual pillars the rotational latency is halved for any given RPM compared to a single pillar design. Therefore dual pillar 3600 rpm ~3" platter is going to have the same rotational latency as a single pillar 7200 rpm 2.5" platter. However, the much lower RPM (of the dual pillar 3.5") will reduce power consumption per TB by a large amount. The only way for 2.5" single pillar multi-actuator could regain an advantage would be to develop a "top loader chassis" like the 84 bay 5U 3.5"....but the advantage regained would only be one of data density.

With that mentioned, one thing to consider regarding dual pillar 3.5" vs. single pillar 3.5" is that if halving the RPM and using the smaller and stiffer ~3" platter allows more platters in a 3.5" housing compared to the higher RPM single pillar 3.5"? If so, then the data density between the dual pillar 3.5" and single pillar 3.5" might not end up being different (or if is different....then not by very much).

 

[cbn]

Found out there is actually a 4U 2P Server with 106 3.5" bays:

https://www.inspursystems.com/product/nf5486m5/

https://www.hpcwire.com/2018/06/27/...w-server-with-worlds-highest-storage-density/

https://www.servethehome.com/simply-huge-106x-3-5in-hard-drive-system-the-inspur-nf5486m5/

This is a 58% increase in density over the 5U enclosure with 84 3.5" bays.

 

[thecoolnessrune]

In overall drive shipments, I personally see 2.5" taking the biggest impact.

3.5" drives will still have the market for bulk cold storage. And that is a *huge* portion of the Cloud market.
M.2 will continue to be in the mobile market in the form of SSDs for laptops and similar.
2.5" is popular in server, but I forsee it being heavily impacted by ruler form factors for SSDs.

So I see current 2.5" based servers moving towards ruler form factors, while 3.5" form factors remain for bulk storage. M.2/3.5" continues in the consumer market.

It will really depend on what server vendors do. Right now, the titans of SSDs are ruling over which proprietary form factor will win, and at the same time most Server vendors make a single model server accommodate either LFF (3.5") or SFF (2.5") layouts by changing the cage and backplane during final construction of the server. Ruler form factors on the other hand require a different server layout to make efficient use of the space.

 

[rchunter]

WD is consolidating on their higher end, WD gold drives are now EOL and they switched the name from HGST He drives to WD DC.

https://www.newegg.com/Product/Product.aspx?item=N82E16822234361

https://nascompares.com/2018/08/01/...lternative-hgst-ultrastar-dc-drive-list-here/

 

[Red Squirrel]

Woah that chassis is sexy. Can just imagine the price tag on that though lol. If they don't show the price or even have an option to buy you know you can't afford it.

 

[aigomorla]

Well there are cheaper ones and smaller ones at that too.

I actually like how drives are configured in that style, and like 45 drives as well.
https://www.45drives.com/products/

I was actually thinking of getting one if and when my FreeNAS server ever dies.
I like how the drives are configured, and would think they get more air to keep them cooler under a non server room conditions.

 

[SimplyComplex]

SSDs are 2.5" form factor

I intend to make the world's first 5.25" ssd, by cracking open 4 dozen 1tb Samsung Evos, and stitching them all together with a soldering iron, duct tape, and prayer.

 

[HutchinsonJC]

You're having fun with your post, I'm sure, but there's already SSDs in a different form factor other than 2.5" and my reasoning for posting the comment is the abundantly available market for SSDs in that form factor (2.5") and many OEM computers shipping with that form factor. Plus a general public's thought process of what an SSD even is... "a hard drive replacement" that "takes the place of a mechanical drive" in basically a literal sense. Which means 2.5" form factor.

Not as a hard rule, but you'd almost have to be a tech enthusiast to be looking at m.4. I have people all the time asking me about hardware or suggestions for their home computer. SSD to them, is 2.5" form factor. That's what they know.

That all aside, the guy that mentioned backup drives some few comments after my post as a big push of the 2.5" form factor, in terms of sales, was able to demonstrate pretty clearly, what the 2.5" market looks like in my opinion. I hadn't considered when I originally wrote my first post in this thread that I personally rotate through 10 2.5" backup drives every business day of the week haha (when I say personally, I mean that my hands are on that process and not that it's for personal use).

Anyway, if you aren't just having fun with your post, please be sure to post your progress

 

[cbn]

Regarding M.2 vs. 2.5" I wonder how much differences in NAND packaging and PCB integration will play a role in the future?

Samsung at the time (Dec 2014) of the 850 EVO only used NAND packages with eight dies each in them:

https://www.anandtech.com/show/8747/samsung-ssd-850-evo-review/2

There are three different PCB designs in the 850 EVO lineup. The 120GB and 250GB models (above) use a tiny PCB with room for two NAND packages (one on each side). Interestingly enough, both use octal-die packages, meaning that the 120GB 850 EVO only has a single 128GB (8*16GB) NAND package. Decoding the part number reveals that the packages are equipped with eight chip enablers (CEs), so a single NAND package is viable since all eight dies can be accessed simultaneously.

The use of octal-die packages is actually true for all capacities. It's an interesting choice nevertheless, but I suspect Samsung's packaging technology is advanced and mature enough that it's more cost efficient to use high die count packages and small PCBs instead of larger PCBs with more and less dense NAND packages.

This, in contrast, to 2.5" SSDs like the April 2016 PNY CS2211 where the packages had only two 15nm 128 Gbit MLC NAND dies each (ie, 2*16GB) and thus were more spread out on the PCB:

https://www.anandtech.com/show/10234/the-pny-cs1311-and-cs2211-ssd-review-mlc-vs-tlc-at-15nm

 

[killster1]

What storage form factor due you think will decrease in usage the most over the next 5 years?

do you think there will be something called nvdimm-p for sale?

this thread is the first ive heard of such storage, i wonder what the cost per gb is if you are hyping it so much?


2.5" is my vote since i have not seen the capacity rise like everything else has. 2.5" has a limit of 5tb? still has strong use in gaming consoles. Hard to know numbers tho.