20 years sounds like a long time for new technologies to emerge , but then again I always underestimates how long old technology keeps floating around. btw What ever happend to the hard disk killer technology from HP they wanted to license out.
never mind found it "In October 2011, the HP team announced the commercial availability of memristor technology within 18 months, as a replacement for Flash, SSD, DRAM and SRAM.[10] Commercial availability of new memory was more recently estimated as 2018.[11] In March 2012, a team of researchers from HRL Laboratories and the University of Michigan announced the first functioning memristor array built on a CMOS chip.[12]" Well that din't went accordign to schedule then xD
3D XPoint is even farther away from replacing HDDs than flash is, though. HDDs will still have a market as long as they're significantly cheaper than solid state disks, and 3D XPoint costs a lot more than flash.
3d XPoint might even help HDDs. You then have CPU cache, RAM, 3D XPoint (instead of ssd) and then a hdd for your movies, music and so forth. In fact even now it makes little sense to store media files on an SSD. Not cost effective.
When I read these proclamations I always suspect there is desperation lurking just below. Sure, they may still gouge platter density through their magic for a time much like Intel does with their chips, but it`s becoming much more costly to do so with every new iteration. Will consumer shell out something like 500 bucks for extremely sophisticated 50TB drive, if he`s effectively living in the damn cloud presently, and 50$ for 1TB midclass SSD seems just about enough?
Hard drives are only relevant because they offer larger storage densities at lower prices. That's their only differentiation. Hard drives are the "Wal-Mart" of storage and it's pretty damn hard to try to be Wal-Mart.
Sure, but who buys drives for servers, etc? SSDs are not so relevant for storage purposes and costs much more, enterprise class SSDs costs A LOT MORE. Remember that DAT tapes are still used effectively today as backup devices.
If you can get a 1 TB SSD for 50$ (that's quite some years away) you'll be able to get a 10+ HD for 50$ as well. Sure, there may also be a 50 TB HDD for 500$.. but that's never a viable consumer alternative and is left for the data centers. People with moderate storage requirements will opt for onyl the SSD, others will go for the SSD and the HDD. Or embark the cloud and have their stuff stored on other peoples HDDs.
I imagine something like a 1TB SSD daily driver and that 500$ 50TB HDD for long-term storage. Still, that`s just one drive instead of many they are selling today.
SSD is not ready for 4k content for a quite a long time. So there is a need for ultra high capacity drives. SSD are developing fast to high capacity drives, but they are not going to be ready for 4k or 8k content for a long time. Normal users are fine with pure ssd in few years, if you are not in very tight budget.
SSDs are far cheaper per gigabyte today than HDDs were when 1080p content became a thing in 2003. Since most 4K video content will be streamed or on optical discs, saying that SSDs aren't ready for 4K content doesn't really make sense... particularly because all modern digital video capture, 4K or otherwise (be it directly on cameras or via recording decks) is done on SSDs.
Its not always SSDs. There are all based on flash memories but not always in SSD form.
4K still makes HDD a necessity. For every 10 SSDs involved in a video production, at least double amount of HDD is used to copy, transport, edit and archive the clips generated from the SSD based recorder.
It sounds that way to me, too. Second largest producer of a commodity product claims their business model is sound for the long term! Stockholders everywhere rejoice!
Still, they're probably right. Hard drives are awfully popular for mass storage and flash isn't cheap enough to dethrone it.
The truth is, they are right though. Consumers may switch to SSD in the next 5 years (already happening), but for every consumer with a phone or a tablet, you will need more storage in the cloud for their media, be it what needs to be streamed to them (like netflix) or what they are producing (home movies, pictures...). For such cold storage (files are rarely accessed and only by a few), HDDs make a lot of sense. The valume of HDDs and who buy them is shifting from consumers to cloud operators, but the market will be here for a long while, despite Flash progresses.
Increasing capacity is welcome but to stay relevant you need to address performance too. Probably with greater density performance will climb also, but additional cashing and ather technology are needed to make up for performance against ssd. I do not want to copy 10TB worth of data for one week time.
I think they have less incentive to address performance in light of solid state drives.
Hard drives have a very skin window where they can increase cost/gb before solid state storage comes into play and completely owns the performance aspect.
Hard drives have been reduced to a one trick pony: storage density.
They need to tie the performance of SSDs? This would be wasting huge amounts of money for a lost cause. It's impossible in the sense that you'd have to be really stupid about SSD design to make that happen. You don't demand your SSD to match main memory speeds for a good reason, don't you?
Keep the HDDs affordable and "fast enough". If you're concerned about sustained transfer speed simply raid a few drives. If you're concerned about random and general suage speeds.. well, there's nothing HDDs can do for you except going to 10k and 15k rpm. Which doesn't make any sense compared to SSDs.
It's not difficult at all. Just make the drive heads asynchronous and strip data across platters and platter surfaces. Currently a 6TB WD Black reads and writes at about 210MB/s, but all the arms move to the same locations per platter while only one of them actually does any reading or writing. You can actually multiply current HDD performance by a factor of 10 by doing this, putting it in the 2GB/s reading/writing space, and that's only with 1 column of arms. A number of drives today have multiple arm columns, but only one acts at any moment in time.
Seagat and WD will be fine for 2 decades easily. IBM demo'ed asynchronous arms almost 5 years ago. When they enter the foray, along with proper data striping, SSDs will lose the performance benefits on all but the highest end (M.2 ultra NVMe top out at 4GB/s R/W).
That will only increase bandwidth. Latency will still be horrible. The cost of the drive will increase lots too (with a decrease in reliability) as you now need more actuators.
If the additional cost was tiny, one of HDD manufacturers would have done it (demo is not the same thing as real product!). It's not and it's borderline impossible to make it viable. It's hard enough to hold one arm over very narrow track. And you are missing the point of why are SSDs so much faster. It's not about sequential speed - they are competitive enough in this area (per dollar). It's about random access and in that area SSDs are 2-3(some even 4) orders of magnitude faster.
You assume that the tracks will be perfectly aligned so that the heads are in the right position on every platter simultaneously. I imagine this might be possible to achieve once, but any change in environmental conditions would probably throw that off.. After that, you have an engineering challenge of conrolling the temperature precisely enough that the platters remain the exact same size and the tracks remain aligned with those of other platters..
For performance you add additional tiers: NAND SSDs, Intel 3D XPoint or other next-gen memory, and DRAM. A system could include DRAM, XPoint, SSDs, disks, and tape if it were designed carefully.
The chart is hard drives SALES, not hard drive capacity. Moving to cloud storage means fewer but larger drives, likely operating closer to capacity. "Increasing usage of cloud storage" is absolutely a factor in shrinking sales numbers.
Cloud storage utilizes large 3+ TB capacity drives, while home PC's typically use 1-2TB hard disks. 3+ TB disks are a niche market for home PC's. This is enough to make the popularity of cloud storage decrease per unit sales.
My family uses as 24TB RAID 10 of WD Blacks for our joint backup server (6TB for each of us). The beautiful thing is having 10Gb NICs in all of our computers that can sustain reading from our 2x256GB RAID 0 SSD setups at full pace (until 1TB M.2 NVMe drives come along).
Full speed locally and plenty of remote speed to send data to its resting place. Oh, and the storage server runs 4x4GB of DDR4 3200MHz, so there's plenty of cache to store stuff in while it's streaming into the HDDs.
I think lots of Mac users have 3 or 4TB drives as Time Machine drives because, why not? A longer archive history may turn out to be useful. If MS ever got their act together enough to ship something as seamless and functional as Time Machine, we might see the sales of large external drives jump substantially.
I'm wondering if this is some of the changes we are going to get as Anandtech's new owners cheapen the site. Anton Shilov is a writer for X-bit Labs? But now getting articles on the front page of Anandtech? It reads more like a brochure from mechanical hard drive companies than a technical article, with grand statements like "HDDs will remain a key bulk storage technology for a long time.", which is supported by absolutely nothing.
This isn't an article, it's a shill piece for Seagate and other mechanical drive makers. It's great that Seagate has all this new technology coming out, but it's putting frosting on a dog turd and everyone knows it. So fricking say it and don't "write" garbage pieces like this. Quit being afraid of your corporate masters, Anandtech.
It's a pipeline article. A lot of them have been press-releasy ever since pipeline was created to replace the dailytech feed.
Anton doesn't appear to've written for Xbit in over a year (if you were thrown by his twitter account coming up in Google; looks like he abandoned that 2 years ago). Until 2 months ago he was writing for Kit-Guru. I'd guess he's Anandtech's newest hire.
"Bulk storage2 doesn't mean it would be in every device, it simply means this technology will store the bulk of the data.. which is obviously valid. Most data we store is accessed so seldomly, it would be downright stupid to store all of it on flash or other costly high performance memory.
I disagree. I want to read about new technology and industry marketing speak isn't going to make it more likely that I purchase HDDs, only $/TB will.
This is hardly all about Seagate since the whole industry will be adopting these technologies and helium-filled drives, an HGST/WD innovation, are mentioned.
x2 there is some real content to this even, explaining the newer technologies. I had never heard of TDMR, BPMR, or HDMR prior to this article. I had only heard of PMR, SMR, and HAMR.
"I'm wondering if this is some of the changes we are going to get as Anandtech's new owners cheapen the site. "
Just so there's no confusion here, the site, its employees, and the content featured here are all under my control. As Dan correctly picks up on, Anton is a new hire for AnandTech, as I wanted to bring on someone to keep the pipeline busier for you guys so that you'd have more new things to read here every day. At the same time Anton wanted to write about hard drives - and specifically Seagate's recent comments - so I let him.
No element of this article comes from our "corporate masters." The writers I choose are writing about the subjects they want to write about (and I approve). We don't do sponsored content or any other system where someone above me tells us what to write. So if you have any comments about the content here - good or bad - I am the one responsible to the readers for the site and its contents.
Yes, we do need that clarification. These days Anandtech's bourgeois repertoire is closer to a cross between tom's hardware and GSM Arena than the Anandtech we remember from the 90's and 2000's. It's not entirely their fault because the authors and audience are, put simply, different now. Enough Rubicons have been crossed such that redemption to loftier realms of content and discussion are not realistic, but the readership is growing. Yes I want some parts of my childhood back, including the stroke of wonder I used to get from a new AT writeup. But that's too bad. You can never go home, but you can always shop there.
For everyone's sake, yes. Let's make it clear that "the cloud" is simply dumbed-down marketable language that represents the trending migration of computer resources away from the user and into a managed, hosted environment. It represents an increase in total HDD shipments, but a decrease in consumer HDD shipments.
The Internet is getting bigger. More and more computers are being built and more hard drives to go inside them, they just aren't being stocked on shelves because they aren't being sold to John Everyman Smith. Somebody figured out they can make more money managing the systems and selling "whatever" as a service that you must buy subscriptions to access.
YOU ARE SO RIGHT. Those are EXACTLY my thoughts quite often when I visit this site. I really liked how articles were written and they had their own style - they used to know some more English (grammar) too. Reading used to be a smooth process of acquiring information and storing it with a certain flow. Reading current articles here feels like trying to solve a Rubik's cube. I have to overprocess sentences and mentally rearrange them. Now, as I'm thinking of it, I believe this site (and the article quality check) has declined considerably since Anand left for CrApple. That must have been the moment things started to fade slowly.
Unfortunately, I see myself reading other websites and enjoying the content much more than the stuff on Anandtech. I used to visit Anandtech every single day, sometimes multiple times a day because they had me addicted. Now it's sadly become a "meh...nah". Comparing past to present, it feels as if not only Anand's gone, but also Ian and some others who, frankly, were really talented at writing articles. Everyone can type on a keyboard, but not everyone can keep readers engaged. I remember those book-size reviews by Anand - I used to open the drop down menu and see tons of pages for a single article and my reaction used to be "damn, this thing is long, and I CAN'T WAIT to read every word of it". They were tech teachers and English teachers at the same time. This has turned into what we call "skimming" on this site. I'm sad to say this site shouldn't bear Anand's name. It's lost the Anand in Anandtech (take it as a pun too). It's just "ANother TECH review site". Maybe Antech would make more sense as a website title, given the current state this place is in. I fully understand it when you say you want parts of your childhood back. The good times have apparently come and gone for this site. I guess sometimes you gotta lose what you have to understand its value.
Consumers have hard drives with large % of unused capacity. "The cloud" datacenters have to use the space they buy efficiently. If consumers aren't storing data locally, that means less shipments of HDDs.
Someone already touched on this, but I think performance is the bigger issue than density, to the consumer. As files become larger and larger (due to higher definition etc.) the need for larger storage may once again be relevant, but I think the biggest issue HDD's face today is performance.
Not only is 4K adoption slow, but the content gatekeepers are trying to keep 4K "locked in the cloud" (Netflix and other streaming services). The consumer doesn't need to store as many video files unless they are a serious pirate or are filming their own content. H.265 reduces the amount of capacity needed to store 4K video and other codecs like Perseus or VP10+ might reduce it even more.
If you want performance, you use an SSD. Hybrid hard drives have not been too successful although they are a good idea going forward. I assume the "Hybrid Enhanced Cache" mentioned in the technology chart is talking about an attempt to revitalize the hybrid hard drive concept.
HDD manufacturers can try to bridge the performance gap, but should remain focused on increasing capacity and lowering cost.
Simply by increasing areal density, the sequential performance goes up as more data flys under the heads per unit time. There is no need to really specifically focus on it anyways nor is there really anything you can do. Higher RPM help a bit, but it impractical for consumer drives, and there is no hope for random performance. Hybrid drives can help but you can see that is already mentioned in this post.
20 years? I highly doubt they will remain relevant as very high-capacity flash (or whatever replaces flash) drives are completely mainstream in 5 years.
I've seen a 2025 estimate for flash overtaking HDDs in TB/$ written earlier this year; that author seems fairly flash friendly; his TCO estimate is that flash is within a year of displacing HDDs for general purpose storage in the data center (where power usage, density, and even being able to pack one more VM/server on servers that already run a dozen or so matters). While Seagate may be overly optimistic on the lingering strength of HDDs, predicting more than a few years out is hard since it depends on how successful the commercialization of tech that's in early stages of the R&D pipeline ends up being.
When cheap SDDs are able to turf out HDDs from $400 boxmart specials also probably matters. My guess would be that we'll see it start taking a largish bite of that market around when 256 or 512GB is available for <$50 (to undercut cheap 2.5" HDDs). While the average customer in that market probably only really needs 64 or 128; those numbers are probably a lot lower than the ones in their old cheap junk PC. Convincing uneducated customers to take something that's "not as good" (because of the smaller number) as their old PC is a tough sell since GB is an easier to understand number than all the SSD performance gains are.
I don't need a lot of storage, and unless I start downloading a bunch of high-quality 4K video, I don't think that's going to change for a while. When my current 1TB HDDs fail, I'm done with HDDs.
Brag all you want about your tech Seagate, but unless the reliability of consumer drives improves, I'm not touching spinning platters anymore. I always have at least one hard-drive failed in less than a year, regardless of its usage (external, internal, desktop, laptop). Thank goodness there's a company like Apple that decided to switch to SSD on all their laptops. I even have SSD as my external storage now since two different external hard drive I bought failed back to back, all in less than 6 months of usage.
The only place I use hard drives now is my drobo, and that's because there's redundancy. No redundancy, no spinning platter for me. I'm not trusting terabytes of data on some mechanical contraption that will fail without warning.
lol are you guys doing satire now? This can't even be called advertising anymore. Relevant in what field? Cold storage, maybe but do we give a damn about that? Demand for devices with HDDs has stalled, the lack of competition has kept price flatish, demand for home storage is weak due to streaming. NAND prices have been dropping 30% or more per year, even in home storage HDDs will be done soon. In enterprise you factor in power and density and not much room for HDDs left. You could have a TB of NAND at 25$ in 5 years and 4TB microSDs. At the International Solid-State Circuits Conference in February Micron will talk about a 768-Gbit (96GB) chip in 179.2mm2. Someone like Seagate will say robots are a big deal for storage but not for HDDs when you factor in cost, power and density. At this point in consumer notebooks a small % have SSDs but that will change fast. To make it worse, flexible screens will hit PC sales really hard soon enough and glasses will end PCs. They are running out of markets and they are too greedy to lower prices enough to remain relevant a bit longer. Hell, at this point it is time to start talking about the demise of NAND. New memories will do to NAND what NAND did to HDDs and that's about to start.
V-NAND has postponed the demise of NAND. Post-NAND technologies have been long promised but have yet to appear. Even Intel's 3D Xpoint will occupy a tier between and alongside DRAM and NAND due to expense.
Hmm. I doubt I will ever buy a computer with a HDD again. The last time I did was five and a half years ago. But for my NAS hard drives (Western Digital Red) make sense to me. Anyway, I thought I had the latest and greatest. A Dell XPS 13 with Skylake and Samsung NVMe SSD. But it turns out Intel and Micron have this technology called 3D XPoint.
Don't get too excited about xpoint just yet, they'll try to milk it as much as they can. Building capacity is rather costly so they won't convert much to it at first. On the upside DRAM prices have dropped plenty recently and xpoint has to be cheaper so we could see 80GB xpoint SSDs at 199$ , if not now then with the second gen xpoint that is supposed to arrive pretty soon. The real fun will be when a bunch of other new memories become available (xpoint seems likely to be a stopgap product).
As for storage, how about in 10 years from now when you have 3D ReRAM at 10$ per TB ( if not 2-3 times cheaper). How much cheaper would HDDs need to be to convince you to buy one? Ofc the average Joe doesn't need all that much home storage and NAND will be affordable for such buyers soon enough.
ReRAM/Crossbar/memristors are empty promises until they hit the market, but once they do appear they could rapidly make up lost ground. NAND has a great chance of killing the hard drive. The only problem the industry has right now is lowering the cost per terabyte. V-NAND is helping to drive prices down. The days when $0.50/GB was a good deal are gone.
If Samsung wanted to, they have the technology to put out a 256 terabyte 2.5" drive (using both V-NAND and TSV stacked dies) TODAY.
Everything else aside, one thing needs to be clarified. Samsung doesn't have NAND with TSV, even less so 3D NAND with TSV where it would be a lot harder to do TSV and rather wasteful (die area wise).
The thing about SSD's is that the cost of the NAND makes the product essentially have no market. Sure they could build a drive with way more capacity, but if they have to sell it at $4k or $8k or something then they will sell so few they would never have a chance to make money on the product.
Just give me a cheap 8TB SSD so I can replace my 3TB HDD's. Doesn't really even need to be faster than a HDD, just cheap, reliable and low power. Surely there is a market other than me for NAS/stprage SSD's? Less power than a HDD, also cooler and silent.
Because of price. Samsung can already fit 4TB of NAND in a 2.5" drive and it will probably cost ~$1500 based on the fact that the 2TB one costs >$800 -- so by making a bigger ssd to fit more flash -- its going to cost several thousand dollars. Who would buy that? Too few people at the moment.
Now in the enterprise world you already see this -- although it is on PCIe cards instead of 3.5" form-factor SSD's. You can buy PCIe cards with 3-4+TB of space, for several thousand dollars and that's a thing in that market, but just wouldn't work for consumer.
They would have SIGNIFICANTLY higher read/write speeds .. in sequential. The 10TB HE drives are already pushing 240-250MB a sec. As the density goes up the sequential speed naturally goes up too as more data is flying past the head per unit time. a 50TB drive should easily be able to do 1-2GB a sec sequentially.
Since the flood of Thailand i noticed the following: - Hard-drive prices took a hike of 300%. I remember buying top of the line hard-drive of 2Tb back then in 2011 for merely $60. Today's price is $180 for the very same drive. - The consumer market is still stuck around 4TB hard-drive size, consumers do not want to shift to higher capacities because the price is too high, it starts doubling for 6 Tb. - Hard-drive prices will continue to be expensive because they have to maintain 2 technologies and despite their shipment they never recovered cheaper prices. - I would wait next year with a sharp decline in hard-drive prices because the global demand will slow down
I can see HDD lasting for some time down the road for mass data center storage and maybe low end budget systems but I would very surprised if HDDs survived for the next 20 years in the consumer space. Any consumers with interest in these super high capacity HDDs is going to be aware of the drawbacks and nobody like that is going to want to store so much data on a single point of failure device with such low performance. What's that? Just use RAID, you say? Sure, then when you get a couple of drive failures you can spend the next six month rebuilding your array of dozens of terabytes of data at 80MB per second. Have fun with that. RAID makes sense for mass scale storage deployments where a company has the time and resources to deal with live rebuilds for enterprise applications but not for consumers. As an end user it's frequently a tremendous frustration for me that my super fast SSD is brought to its knees while waiting for media transfers to and from my HDD at 80-120MB/s. A system can only be as fast as its slowest component.
As for 3D Xpoint, it's a 3D stackable technology, so according to Intel it's not going to be that difficult to ramp up capacity. The technology is already in production and Intel has stated that devices for consumers will be available in 2016. Personally, I'm waiting to see what happens in 2016 with this technology before upgrading or any new drives.
Another point I've not seen addressed here is that a lot of these spiffy new technologies being developed to stretch HDD capacities are going to increase the cost of the drives. They are improving capacities and increasing production costs while retaining all of the weaknesses of current HDD technology.
How about crinkle-cut platters so they have more surface area?
I think one way or another the long-term future of storage is going to be 3D, either by stacking NAND, or holographic. It's a shame they don't know how to rewrite magnetic domains in the interior of the platter (depthwise).
I dont think any consumers, or even geek cares about the HDD capacity anymore. The Consumer HDD drive business, much like the PC market is shrinking. But similar to Intel's case the growing appetite in Cloud Business has increased more then enough to cover the the loses.
HDD will continue to thrive as long as we continue to create huge amount of Data. NAND, even with TSV is not going to take over HDD in the next 10 years. I am not sure about 20. But definitely not 10. Some people will go about saying Micro SD went from 128MB to 128GB in 10 years, but that is forgetting NAND went from using many node generation behind leading node to now basically they are the leading node to test. The headroom in 2003 were huge, and MLC were being tested as well as TLC. Today we have TSV that could stack to unknown number of layers. I remember reading an engineers saying there are nothing stopping them to do 1024 layers, they dont know what sort of problem they would face when it is that many layers.
But even at TSV 1024 layers, the cost per TB still doesn't rival HDD. Not to mention the power / TB also flavors HDD.
So definitely not in 10 years. 20years? I dont know since that really is too far fetched, but the 10 year roadmap for both technology has pretty much set in stone.
Some will also say the price for Cloud Storage will drop to very very low, an leading example would be Blackbaze B2. However, the cost of GB % has shrinking to the point where it is now largely dominated by Bandwidth cost and Power cost. And both cost has not even shrink in MANY years. It is a bit like Amazon AWS continue to give you more Memory and CPU resources, but your port speed and bandwidth transfer stays the same.
How does increased use of the cloud decrease the amount of HDD space needed? Does the cloud just have a really good memory and not need to write stuff down?
Not to mention deduplication. For example Google and Apple don't store every song you "upload" to their cloud indiviually. And better compression. Though to a lesser degree as most big files are already compressed rather well (pictures, video, ...).
I can definitely see HDDs being relevant IF (big if) they keep ahead on $/GB. Currently flash is around what 40c? Per GB. HDDs are at around 6-8c. I wish I could have my Nas run ZFS on SSD... But somehow I don't think it is going to happen for a long while, if at all. And as time goes on and capacity needs expand.. So will the inability of SSD's to store data. By all means kill HDDs in your PCs. But large storage is HDD bound and will be for some time.
I think HD might still be relevant if they can reach "good enough" performance coupled with the areal density advantage. Just in the consumer space, we have to remember we don't represent the majority. I was hell bent on upgrading everyone in my family to SSD. An aunt running an i5 4200U laptop with 750GB 5400RPM HD (stock) is the lone holdout, and I was about to do it this Thanksgiving but then she insisted what she has is plenty fast. I observed her usage over the break and honestly, a SSD would've given her awesome startup and program launch time, but that's it. And Windows 10 has come a long way towards reducing boot time even on HD. I ended up not pushing the matter, since at the end it's her laptop, but also because I actually didn't feel too guilty about her current state - again given her usage pattern, which I suspect mirrors a decent portion of the general consumer.
It is really amazing article on <a href="http://www.zeepedia.com/answers/question-category/... Drive/">Hard Drive</a>. I could not see such article ever before. The method of explaining hard drive and its various parts is very amazing. This article contain very useful information.
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
86 Comments
Back to Article
plopke - Friday, December 18, 2015 - link
20 years sounds like a long time for new technologies to emerge , but then again I always underestimates how long old technology keeps floating around.btw What ever happend to the hard disk killer technology from HP they wanted to license out.
plopke - Friday, December 18, 2015 - link
never mind found it "In October 2011, the HP team announced the commercial availability of memristor technology within 18 months, as a replacement for Flash, SSD, DRAM and SRAM.[10] Commercial availability of new memory was more recently estimated as 2018.[11] In March 2012, a team of researchers from HRL Laboratories and the University of Michigan announced the first functioning memristor array built on a CMOS chip.[12]" Well that din't went accordign to schedule then xDhttps://en.wikipedia.org/wiki/Memristor
Michael Bay - Friday, December 18, 2015 - link
Memristor is dead anyway, HP diverted funding into something closer to HBM.nandnandnand - Friday, December 18, 2015 - link
Intel's 3D XPoint is close to what was promised with memristors and Crossbar RRAM.Guspaz - Friday, December 18, 2015 - link
3D XPoint is even farther away from replacing HDDs than flash is, though. HDDs will still have a market as long as they're significantly cheaper than solid state disks, and 3D XPoint costs a lot more than flash.beginner99 - Saturday, December 19, 2015 - link
3d XPoint might even help HDDs. You then have CPU cache, RAM, 3D XPoint (instead of ssd) and then a hdd for your movies, music and so forth. In fact even now it makes little sense to store media files on an SSD. Not cost effective.MrPoletski - Monday, December 28, 2015 - link
Me? I always underestimate how quickly 20 years goes by.Michael Bay - Friday, December 18, 2015 - link
When I read these proclamations I always suspect there is desperation lurking just below.Sure, they may still gouge platter density through their magic for a time much like Intel does with their chips, but it`s becoming much more costly to do so with every new iteration.
Will consumer shell out something like 500 bucks for extremely sophisticated 50TB drive, if he`s effectively living in the damn cloud presently, and 50$ for 1TB midclass SSD seems just about enough?
ImSpartacus - Friday, December 18, 2015 - link
There is definitely desperation.Hard drives are only relevant because they offer larger storage densities at lower prices. That's their only differentiation. Hard drives are the "Wal-Mart" of storage and it's pretty damn hard to try to be Wal-Mart.
sfuzzz - Friday, December 18, 2015 - link
Sure, but who buys drives for servers, etc? SSDs are not so relevant for storage purposes and costs much more, enterprise class SSDs costs A LOT MORE. Remember that DAT tapes are still used effectively today as backup devices.Michael Bay - Friday, December 18, 2015 - link
I suspect tapes will outlive HDDs purely on cost basis.MrSpadge - Friday, December 18, 2015 - link
If you can get a 1 TB SSD for 50$ (that's quite some years away) you'll be able to get a 10+ HD for 50$ as well. Sure, there may also be a 50 TB HDD for 500$.. but that's never a viable consumer alternative and is left for the data centers. People with moderate storage requirements will opt for onyl the SSD, others will go for the SSD and the HDD. Or embark the cloud and have their stuff stored on other peoples HDDs.Michael Bay - Friday, December 18, 2015 - link
I imagine something like a 1TB SSD daily driver and that 500$ 50TB HDD for long-term storage.Still, that`s just one drive instead of many they are selling today.
haukionkannel - Friday, December 18, 2015 - link
SSD is not ready for 4k content for a quite a long time. So there is a need for ultra high capacity drives. SSD are developing fast to high capacity drives, but they are not going to be ready for 4k or 8k content for a long time. Normal users are fine with pure ssd in few years, if you are not in very tight budget.Guspaz - Friday, December 18, 2015 - link
SSDs are far cheaper per gigabyte today than HDDs were when 1080p content became a thing in 2003. Since most 4K video content will be streamed or on optical discs, saying that SSDs aren't ready for 4K content doesn't really make sense... particularly because all modern digital video capture, 4K or otherwise (be it directly on cameras or via recording decks) is done on SSDs.crimsonson - Saturday, December 19, 2015 - link
Its not always SSDs. There are all based on flash memories but not always in SSD form.4K still makes HDD a necessity. For every 10 SSDs involved in a video production, at least double amount of HDD is used to copy, transport, edit and archive the clips generated from the SSD based recorder.
Mr Perfect - Friday, December 18, 2015 - link
It sounds that way to me, too. Second largest producer of a commodity product claims their business model is sound for the long term! Stockholders everywhere rejoice!Still, they're probably right. Hard drives are awfully popular for mass storage and flash isn't cheap enough to dethrone it.
frenchy_2001 - Tuesday, December 22, 2015 - link
The truth is, they are right though.Consumers may switch to SSD in the next 5 years (already happening), but for every consumer with a phone or a tablet, you will need more storage in the cloud for their media, be it what needs to be streamed to them (like netflix) or what they are producing (home movies, pictures...).
For such cold storage (files are rarely accessed and only by a few), HDDs make a lot of sense.
The valume of HDDs and who buy them is shifting from consumers to cloud operators, but the market will be here for a long while, despite Flash progresses.
Alien959 - Friday, December 18, 2015 - link
Increasing capacity is welcome but to stay relevant you need to address performance too. Probably with greater density performance will climb also, but additional cashing and ather technology are needed to make up for performance against ssd. I do not want to copy 10TB worth of data for one week time.ImSpartacus - Friday, December 18, 2015 - link
I think they have less incentive to address performance in light of solid state drives.Hard drives have a very skin window where they can increase cost/gb before solid state storage comes into play and completely owns the performance aspect.
Hard drives have been reduced to a one trick pony: storage density.
MrSpadge - Friday, December 18, 2015 - link
They need to tie the performance of SSDs? This would be wasting huge amounts of money for a lost cause. It's impossible in the sense that you'd have to be really stupid about SSD design to make that happen. You don't demand your SSD to match main memory speeds for a good reason, don't you?Keep the HDDs affordable and "fast enough". If you're concerned about sustained transfer speed simply raid a few drives. If you're concerned about random and general suage speeds.. well, there's nothing HDDs can do for you except going to 10k and 15k rpm. Which doesn't make any sense compared to SSDs.
patrickjp93 - Friday, December 18, 2015 - link
It's not difficult at all. Just make the drive heads asynchronous and strip data across platters and platter surfaces. Currently a 6TB WD Black reads and writes at about 210MB/s, but all the arms move to the same locations per platter while only one of them actually does any reading or writing. You can actually multiply current HDD performance by a factor of 10 by doing this, putting it in the 2GB/s reading/writing space, and that's only with 1 column of arms. A number of drives today have multiple arm columns, but only one acts at any moment in time.Seagat and WD will be fine for 2 decades easily. IBM demo'ed asynchronous arms almost 5 years ago. When they enter the foray, along with proper data striping, SSDs will lose the performance benefits on all but the highest end (M.2 ultra NVMe top out at 4GB/s R/W).
The additional engineering cost is tiny.
Gigaplex - Friday, December 18, 2015 - link
That will only increase bandwidth. Latency will still be horrible. The cost of the drive will increase lots too (with a decrease in reliability) as you now need more actuators.qap - Friday, December 18, 2015 - link
If the additional cost was tiny, one of HDD manufacturers would have done it (demo is not the same thing as real product!). It's not and it's borderline impossible to make it viable. It's hard enough to hold one arm over very narrow track.And you are missing the point of why are SSDs so much faster. It's not about sequential speed - they are competitive enough in this area (per dollar). It's about random access and in that area SSDs are 2-3(some even 4) orders of magnitude faster.
saratoga4 - Sunday, December 20, 2015 - link
>It's not difficult at all. Just make the drive heads asynchronous and strip data across platters and platter surfaces.That's all you have to do!
shadowjk - Monday, December 28, 2015 - link
You assume that the tracks will be perfectly aligned so that the heads are in the right position on every platter simultaneously. I imagine this might be possible to achieve once, but any change in environmental conditions would probably throw that off.. After that, you have an engineering challenge of conrolling the temperature precisely enough that the platters remain the exact same size and the tracks remain aligned with those of other platters..nandnandnand - Friday, December 18, 2015 - link
For performance you add additional tiers: NAND SSDs, Intel 3D XPoint or other next-gen memory, and DRAM. A system could include DRAM, XPoint, SSDs, disks, and tape if it were designed carefully.e_sandrs - Friday, December 18, 2015 - link
This always kills me: "Shipments of HDDs decrease due to a variety of factors nowadays, including...increasing usage of cloud storage"Oh, ok, now that data is being stored "in the cloud", I'm glad we don't need HDDs anymore! [/sarcasm]
Psst: There is NO CLOUD, just other people's computers. Do we need that clarification at Anandtech today?
Sunrise089 - Friday, December 18, 2015 - link
The chart is hard drives SALES, not hard drive capacity. Moving to cloud storage means fewer but larger drives, likely operating closer to capacity. "Increasing usage of cloud storage" is absolutely a factor in shrinking sales numbers.MrSpadge - Friday, December 18, 2015 - link
Yep. And it's also a factor in increasing the average selling price of HDDs.lehtv - Friday, December 18, 2015 - link
Cloud storage utilizes large 3+ TB capacity drives, while home PC's typically use 1-2TB hard disks. 3+ TB disks are a niche market for home PC's. This is enough to make the popularity of cloud storage decrease per unit sales.patrickjp93 - Friday, December 18, 2015 - link
My family uses as 24TB RAID 10 of WD Blacks for our joint backup server (6TB for each of us). The beautiful thing is having 10Gb NICs in all of our computers that can sustain reading from our 2x256GB RAID 0 SSD setups at full pace (until 1TB M.2 NVMe drives come along).Full speed locally and plenty of remote speed to send data to its resting place. Oh, and the storage server runs 4x4GB of DDR4 3200MHz, so there's plenty of cache to store stuff in while it's streaming into the HDDs.
patrickjp93 - Friday, December 18, 2015 - link
That should have read as 4x256GB SSD RAID 0 setups.extide - Friday, December 18, 2015 - link
Parent post is still correct. 3+TB at homes is niche. I have a ~big ZFS array too, but I don't let myself think that is common either.extide - Friday, December 18, 2015 - link
I mean, of course, lehtv's post is correct.name99 - Friday, December 18, 2015 - link
I think lots of Mac users have 3 or 4TB drives as Time Machine drives because, why not? A longer archive history may turn out to be useful.If MS ever got their act together enough to ship something as seamless and functional as Time Machine, we might see the sales of large external drives jump substantially.
Phiro69 - Friday, December 18, 2015 - link
I'm wondering if this is some of the changes we are going to get as Anandtech's new owners cheapen the site. Anton Shilov is a writer for X-bit Labs? But now getting articles on the front page of Anandtech?It reads more like a brochure from mechanical hard drive companies than a technical article, with grand statements like "HDDs will remain a key bulk storage technology for a long time.", which is supported by absolutely nothing.
This isn't an article, it's a shill piece for Seagate and other mechanical drive makers. It's great that Seagate has all this new technology coming out, but it's putting frosting on a dog turd and everyone knows it. So fricking say it and don't "write" garbage pieces like this. Quit being afraid of your corporate masters, Anandtech.
DanNeely - Friday, December 18, 2015 - link
It's a pipeline article. A lot of them have been press-releasy ever since pipeline was created to replace the dailytech feed.Anton doesn't appear to've written for Xbit in over a year (if you were thrown by his twitter account coming up in Google; looks like he abandoned that 2 years ago). Until 2 months ago he was writing for Kit-Guru. I'd guess he's Anandtech's newest hire.
MrSpadge - Friday, December 18, 2015 - link
"Bulk storage2 doesn't mean it would be in every device, it simply means this technology will store the bulk of the data.. which is obviously valid. Most data we store is accessed so seldomly, it would be downright stupid to store all of it on flash or other costly high performance memory.nandnandnand - Friday, December 18, 2015 - link
I disagree. I want to read about new technology and industry marketing speak isn't going to make it more likely that I purchase HDDs, only $/TB will.This is hardly all about Seagate since the whole industry will be adopting these technologies and helium-filled drives, an HGST/WD innovation, are mentioned.
extide - Friday, December 18, 2015 - link
x2 there is some real content to this even, explaining the newer technologies. I had never heard of TDMR, BPMR, or HDMR prior to this article. I had only heard of PMR, SMR, and HAMR.More content is always good.
Ryan Smith - Friday, December 18, 2015 - link
"I'm wondering if this is some of the changes we are going to get as Anandtech's new owners cheapen the site. "Just so there's no confusion here, the site, its employees, and the content featured here are all under my control. As Dan correctly picks up on, Anton is a new hire for AnandTech, as I wanted to bring on someone to keep the pipeline busier for you guys so that you'd have more new things to read here every day. At the same time Anton wanted to write about hard drives - and specifically Seagate's recent comments - so I let him.
No element of this article comes from our "corporate masters." The writers I choose are writing about the subjects they want to write about (and I approve). We don't do sponsored content or any other system where someone above me tells us what to write. So if you have any comments about the content here - good or bad - I am the one responsible to the readers for the site and its contents.
TekBoi - Friday, December 18, 2015 - link
Thank you sir.Impulses - Saturday, December 19, 2015 - link
More pipeline, yay.alyarb - Friday, December 18, 2015 - link
Yes, we do need that clarification. These days Anandtech's bourgeois repertoire is closer to a cross between tom's hardware and GSM Arena than the Anandtech we remember from the 90's and 2000's. It's not entirely their fault because the authors and audience are, put simply, different now. Enough Rubicons have been crossed such that redemption to loftier realms of content and discussion are not realistic, but the readership is growing. Yes I want some parts of my childhood back, including the stroke of wonder I used to get from a new AT writeup. But that's too bad. You can never go home, but you can always shop there.For everyone's sake, yes. Let's make it clear that "the cloud" is simply dumbed-down marketable language that represents the trending migration of computer resources away from the user and into a managed, hosted environment. It represents an increase in total HDD shipments, but a decrease in consumer HDD shipments.
The Internet is getting bigger. More and more computers are being built and more hard drives to go inside them, they just aren't being stocked on shelves because they aren't being sold to John Everyman Smith. Somebody figured out they can make more money managing the systems and selling "whatever" as a service that you must buy subscriptions to access.
3ogdy - Saturday, December 19, 2015 - link
YOU ARE SO RIGHT. Those are EXACTLY my thoughts quite often when I visit this site.I really liked how articles were written and they had their own style - they used to know some more English (grammar) too. Reading used to be a smooth process of acquiring information and storing it with a certain flow. Reading current articles here feels like trying to solve a Rubik's cube. I have to overprocess sentences and mentally rearrange them.
Now, as I'm thinking of it, I believe this site (and the article quality check) has declined considerably since Anand left for CrApple. That must have been the moment things started to fade slowly.
Unfortunately, I see myself reading other websites and enjoying the content much more than the stuff on Anandtech. I used to visit Anandtech every single day, sometimes multiple times a day because they had me addicted. Now it's sadly become a "meh...nah".
Comparing past to present, it feels as if not only Anand's gone, but also Ian and some others who, frankly, were really talented at writing articles. Everyone can type on a keyboard, but not everyone can keep readers engaged. I remember those book-size reviews by Anand - I used to open the drop down menu and see tons of pages for a single article and my reaction used to be "damn, this thing is long, and I CAN'T WAIT to read every word of it". They were tech teachers and English teachers at the same time. This has turned into what we call "skimming" on this site. I'm sad to say this site shouldn't bear Anand's name. It's lost the Anand in Anandtech (take it as a pun too). It's just "ANother TECH review site". Maybe Antech would make more sense as a website title, given the current state this place is in.
I fully understand it when you say you want parts of your childhood back. The good times have apparently come and gone for this site. I guess sometimes you gotta lose what you have to understand its value.
jospoortvliet - Sunday, December 20, 2015 - link
Pfff, people will always complain. Still many anandtech pipeline articles are bigger and more thorough than full reviews on most other sites...nandnandnand - Friday, December 18, 2015 - link
Consumers have hard drives with large % of unused capacity. "The cloud" datacenters have to use the space they buy efficiently. If consumers aren't storing data locally, that means less shipments of HDDs.Zan Lynx - Friday, December 18, 2015 - link
Yes it seems people often forget that The Cloud stores data on hard drives.fanofanand - Friday, December 18, 2015 - link
Someone already touched on this, but I think performance is the bigger issue than density, to the consumer. As files become larger and larger (due to higher definition etc.) the need for larger storage may once again be relevant, but I think the biggest issue HDD's face today is performance.nandnandnand - Friday, December 18, 2015 - link
Not only is 4K adoption slow, but the content gatekeepers are trying to keep 4K "locked in the cloud" (Netflix and other streaming services). The consumer doesn't need to store as many video files unless they are a serious pirate or are filming their own content. H.265 reduces the amount of capacity needed to store 4K video and other codecs like Perseus or VP10+ might reduce it even more.If you want performance, you use an SSD. Hybrid hard drives have not been too successful although they are a good idea going forward. I assume the "Hybrid Enhanced Cache" mentioned in the technology chart is talking about an attempt to revitalize the hybrid hard drive concept.
HDD manufacturers can try to bridge the performance gap, but should remain focused on increasing capacity and lowering cost.
extide - Friday, December 18, 2015 - link
Simply by increasing areal density, the sequential performance goes up as more data flys under the heads per unit time. There is no need to really specifically focus on it anyways nor is there really anything you can do. Higher RPM help a bit, but it impractical for consumer drives, and there is no hope for random performance. Hybrid drives can help but you can see that is already mentioned in this post.nathanddrews - Friday, December 18, 2015 - link
20 years? I highly doubt they will remain relevant as very high-capacity flash (or whatever replaces flash) drives are completely mainstream in 5 years.DanNeely - Friday, December 18, 2015 - link
I've seen a 2025 estimate for flash overtaking HDDs in TB/$ written earlier this year; that author seems fairly flash friendly; his TCO estimate is that flash is within a year of displacing HDDs for general purpose storage in the data center (where power usage, density, and even being able to pack one more VM/server on servers that already run a dozen or so matters). While Seagate may be overly optimistic on the lingering strength of HDDs, predicting more than a few years out is hard since it depends on how successful the commercialization of tech that's in early stages of the R&D pipeline ends up being.When cheap SDDs are able to turf out HDDs from $400 boxmart specials also probably matters. My guess would be that we'll see it start taking a largish bite of that market around when 256 or 512GB is available for <$50 (to undercut cheap 2.5" HDDs). While the average customer in that market probably only really needs 64 or 128; those numbers are probably a lot lower than the ones in their old cheap junk PC. Convincing uneducated customers to take something that's "not as good" (because of the smaller number) as their old PC is a tough sell since GB is an easier to understand number than all the SSD performance gains are.
abrowne1993 - Friday, December 18, 2015 - link
I don't need a lot of storage, and unless I start downloading a bunch of high-quality 4K video, I don't think that's going to change for a while. When my current 1TB HDDs fail, I'm done with HDDs.pika2000 - Friday, December 18, 2015 - link
Brag all you want about your tech Seagate, but unless the reliability of consumer drives improves, I'm not touching spinning platters anymore. I always have at least one hard-drive failed in less than a year, regardless of its usage (external, internal, desktop, laptop). Thank goodness there's a company like Apple that decided to switch to SSD on all their laptops. I even have SSD as my external storage now since two different external hard drive I bought failed back to back, all in less than 6 months of usage.The only place I use hard drives now is my drobo, and that's because there's redundancy. No redundancy, no spinning platter for me. I'm not trusting terabytes of data on some mechanical contraption that will fail without warning.
phoenix_rizzen - Friday, December 18, 2015 - link
Do SSDs support any kind of monitoring, like SMART for HDs?extide - Friday, December 18, 2015 - link
Yes, they support SMARTextide - Friday, December 18, 2015 - link
It's not just seagate, fact is they ALL suck, and yes you NEED to store stuff on spinners in a redundant fashion.jjj - Friday, December 18, 2015 - link
lol are you guys doing satire now? This can't even be called advertising anymore.Relevant in what field? Cold storage, maybe but do we give a damn about that?
Demand for devices with HDDs has stalled, the lack of competition has kept price flatish, demand for home storage is weak due to streaming.
NAND prices have been dropping 30% or more per year, even in home storage HDDs will be done soon. In enterprise you factor in power and density and not much room for HDDs left.
You could have a TB of NAND at 25$ in 5 years and 4TB microSDs.
At the International Solid-State Circuits Conference in February Micron will talk about a 768-Gbit (96GB) chip in 179.2mm2.
Someone like Seagate will say robots are a big deal for storage but not for HDDs when you factor in cost, power and density.
At this point in consumer notebooks a small % have SSDs but that will change fast. To make it worse, flexible screens will hit PC sales really hard soon enough and glasses will end PCs.
They are running out of markets and they are too greedy to lower prices enough to remain relevant a bit longer.
Hell, at this point it is time to start talking about the demise of NAND. New memories will do to NAND what NAND did to HDDs and that's about to start.
nandnandnand - Friday, December 18, 2015 - link
People want to hear about new HDD technologies.V-NAND has postponed the demise of NAND. Post-NAND technologies have been long promised but have yet to appear. Even Intel's 3D Xpoint will occupy a tier between and alongside DRAM and NAND due to expense.
JimFoster - Friday, December 18, 2015 - link
Hmm. I doubt I will ever buy a computer with a HDD again. The last time I did was five and a half years ago. But for my NAS hard drives (Western Digital Red) make sense to me.Anyway, I thought I had the latest and greatest. A Dell XPS 13 with Skylake and Samsung NVMe SSD. But it turns out Intel and Micron have this technology called 3D XPoint.
jjj - Friday, December 18, 2015 - link
Don't get too excited about xpoint just yet, they'll try to milk it as much as they can. Building capacity is rather costly so they won't convert much to it at first. On the upside DRAM prices have dropped plenty recently and xpoint has to be cheaper so we could see 80GB xpoint SSDs at 199$ , if not now then with the second gen xpoint that is supposed to arrive pretty soon.The real fun will be when a bunch of other new memories become available (xpoint seems likely to be a stopgap product).
jjj - Friday, December 18, 2015 - link
As for storage, how about in 10 years from now when you have 3D ReRAM at 10$ per TB ( if not 2-3 times cheaper). How much cheaper would HDDs need to be to convince you to buy one? Ofc the average Joe doesn't need all that much home storage and NAND will be affordable for such buyers soon enough.nandnandnand - Friday, December 18, 2015 - link
ReRAM/Crossbar/memristors are empty promises until they hit the market, but once they do appear they could rapidly make up lost ground. NAND has a great chance of killing the hard drive. The only problem the industry has right now is lowering the cost per terabyte. V-NAND is helping to drive prices down. The days when $0.50/GB was a good deal are gone.If Samsung wanted to, they have the technology to put out a 256 terabyte 2.5" drive (using both V-NAND and TSV stacked dies) TODAY.
jjj - Friday, December 18, 2015 - link
Everything else aside, one thing needs to be clarified.Samsung doesn't have NAND with TSV, even less so 3D NAND with TSV where it would be a lot harder to do TSV and rather wasteful (die area wise).
extide - Friday, December 18, 2015 - link
The thing about SSD's is that the cost of the NAND makes the product essentially have no market. Sure they could build a drive with way more capacity, but if they have to sell it at $4k or $8k or something then they will sell so few they would never have a chance to make money on the product.rhangman - Saturday, December 19, 2015 - link
Just give me a cheap 8TB SSD so I can replace my 3TB HDD's. Doesn't really even need to be faster than a HDD, just cheap, reliable and low power. Surely there is a market other than me for NAS/stprage SSD's? Less power than a HDD, also cooler and silent.Hulk - Friday, December 18, 2015 - link
Hard disks are cheap/GB and have fantastic retention. This will make them useful for quite some time.mikegrok - Friday, December 18, 2015 - link
I want a 5 1/4 full height drivephoenix_rizzen - Friday, December 18, 2015 - link
I want a 3.5" SSD. Why limit SSDs to the 2.5" form-factor?extide - Friday, December 18, 2015 - link
Because of price. Samsung can already fit 4TB of NAND in a 2.5" drive and it will probably cost ~$1500 based on the fact that the 2TB one costs >$800 -- so by making a bigger ssd to fit more flash -- its going to cost several thousand dollars. Who would buy that? Too few people at the moment.Now in the enterprise world you already see this -- although it is on PCIe cards instead of 3.5" form-factor SSD's. You can buy PCIe cards with 3-4+TB of space, for several thousand dollars and that's a thing in that market, but just wouldn't work for consumer.
atcronin - Friday, December 18, 2015 - link
I look forward to scrubbing/resilvering a pool containing 50TB drives with 150MB/s read speeds.extide - Friday, December 18, 2015 - link
They would have SIGNIFICANTLY higher read/write speeds .. in sequential. The 10TB HE drives are already pushing 240-250MB a sec. As the density goes up the sequential speed naturally goes up too as more data is flying past the head per unit time. a 50TB drive should easily be able to do 1-2GB a sec sequentially.stevenrix - Friday, December 18, 2015 - link
Since the flood of Thailand i noticed the following:- Hard-drive prices took a hike of 300%. I remember buying top of the line hard-drive of 2Tb back then in 2011 for merely $60. Today's price is $180 for the very same drive.
- The consumer market is still stuck around 4TB hard-drive size, consumers do not want to shift to higher capacities because the price is too high, it starts doubling for 6 Tb.
- Hard-drive prices will continue to be expensive because they have to maintain 2 technologies and despite their shipment they never recovered cheaper prices.
- I would wait next year with a sharp decline in hard-drive prices because the global demand will slow down
TekBoi - Friday, December 18, 2015 - link
Exactly this.Magichands8 - Friday, December 18, 2015 - link
I can see HDD lasting for some time down the road for mass data center storage and maybe low end budget systems but I would very surprised if HDDs survived for the next 20 years in the consumer space. Any consumers with interest in these super high capacity HDDs is going to be aware of the drawbacks and nobody like that is going to want to store so much data on a single point of failure device with such low performance. What's that? Just use RAID, you say? Sure, then when you get a couple of drive failures you can spend the next six month rebuilding your array of dozens of terabytes of data at 80MB per second. Have fun with that. RAID makes sense for mass scale storage deployments where a company has the time and resources to deal with live rebuilds for enterprise applications but not for consumers. As an end user it's frequently a tremendous frustration for me that my super fast SSD is brought to its knees while waiting for media transfers to and from my HDD at 80-120MB/s. A system can only be as fast as its slowest component.As for 3D Xpoint, it's a 3D stackable technology, so according to Intel it's not going to be that difficult to ramp up capacity. The technology is already in production and Intel has stated that devices for consumers will be available in 2016. Personally, I'm waiting to see what happens in 2016 with this technology before upgrading or any new drives.
Another point I've not seen addressed here is that a lot of these spiffy new technologies being developed to stretch HDD capacities are going to increase the cost of the drives. They are improving capacities and increasing production costs while retaining all of the weaknesses of current HDD technology.
stephenbrooks - Friday, December 18, 2015 - link
How about crinkle-cut platters so they have more surface area?I think one way or another the long-term future of storage is going to be 3D, either by stacking NAND, or holographic. It's a shame they don't know how to rewrite magnetic domains in the interior of the platter (depthwise).
iwod - Friday, December 18, 2015 - link
I dont think any consumers, or even geek cares about the HDD capacity anymore. The Consumer HDD drive business, much like the PC market is shrinking. But similar to Intel's case the growing appetite in Cloud Business has increased more then enough to cover the the loses.HDD will continue to thrive as long as we continue to create huge amount of Data. NAND, even with TSV is not going to take over HDD in the next 10 years. I am not sure about 20. But definitely not 10. Some people will go about saying Micro SD went from 128MB to 128GB in 10 years, but that is forgetting NAND went from using many node generation behind leading node to now basically they are the leading node to test. The headroom in 2003 were huge, and MLC were being tested as well as TLC. Today we have TSV that could stack to unknown number of layers. I remember reading an engineers saying there are nothing stopping them to do 1024 layers, they dont know what sort of problem they would face when it is that many layers.
But even at TSV 1024 layers, the cost per TB still doesn't rival HDD. Not to mention the power / TB also flavors HDD.
So definitely not in 10 years. 20years? I dont know since that really is too far fetched, but the 10 year roadmap for both technology has pretty much set in stone.
Some will also say the price for Cloud Storage will drop to very very low, an leading example would be Blackbaze B2. However, the cost of GB % has shrinking to the point where it is now largely dominated by Bandwidth cost and Power cost. And both cost has not even shrink in MANY years.
It is a bit like Amazon AWS continue to give you more Memory and CPU resources, but your port speed and bandwidth transfer stays the same.
Arbie - Friday, December 18, 2015 - link
How does increased use of the cloud decrease the amount of HDD space needed? Does the cloud just have a really good memory and not need to write stuff down?TekBoi - Friday, December 18, 2015 - link
Fewer drives, at larger sizes, operating at near full capacity.simonpschmitt - Saturday, December 19, 2015 - link
Not to mention deduplication. For example Google and Apple don't store every song you "upload" to their cloud indiviually.And better compression. Though to a lesser degree as most big files are already compressed rather well (pictures, video, ...).
doggface - Saturday, December 19, 2015 - link
It amazes me the optimism here.I can definitely see HDDs being relevant IF (big if) they keep ahead on $/GB. Currently flash is around what 40c? Per GB. HDDs are at around 6-8c. I wish I could have my Nas run ZFS on SSD... But somehow I don't think it is going to happen for a long while, if at all. And as time goes on and capacity needs expand.. So will the inability of SSD's to store data. By all means kill HDDs in your PCs. But large storage is HDD bound and will be for some time.
In that light.. Hamr looks interesting.
DaveBG - Saturday, December 19, 2015 - link
In 2025 SSDs will have reached 100TB as well. For the same price so better aim for 200TB.deskjob - Sunday, December 20, 2015 - link
I think HD might still be relevant if they can reach "good enough" performance coupled with the areal density advantage. Just in the consumer space, we have to remember we don't represent the majority. I was hell bent on upgrading everyone in my family to SSD. An aunt running an i5 4200U laptop with 750GB 5400RPM HD (stock) is the lone holdout, and I was about to do it this Thanksgiving but then she insisted what she has is plenty fast. I observed her usage over the break and honestly, a SSD would've given her awesome startup and program launch time, but that's it. And Windows 10 has come a long way towards reducing boot time even on HD. I ended up not pushing the matter, since at the end it's her laptop, but also because I actually didn't feel too guilty about her current state - again given her usage pattern, which I suspect mirrors a decent portion of the general consumer.Philipjohn - Sunday, July 3, 2016 - link
It is really amazing article on <a href="http://www.zeepedia.com/answers/question-category/... Drive/">Hard Drive</a>. I could not see such article ever before. The method of explaining hard drive and its various parts is very amazing. This article contain very useful information.