Hitachi Data Systems to spin of storage division

[LiamC]

via IPO.

http://www.xbitlabs.com/news/storage/display/20101107132159.html

Wonder how this will affect their R&D, like microwave assisted magnetic recording...
http://www.xbitlabs.com/news/storage/display/20101104234947_Hitachi_and_NEDO_to_Co_Develop_Micr owave_Assisted_Magnetic_Recording_for_HDDs.html

[jtr1962]

Translation: Hitachi sees magnetic storage as a dying technology and wants to sell off its disk division while they can still get something for it. In ten years time tops, magnetic disks will be as obsolete as CRTs are now.

[sechs]

Frankly, I had expected that the spinning disk manufacturers would merge or joint venture themselves out of existence.

[LunarMist]

So whatever happened to that 4Tb Hibachi drive?

[Chewy509]

Frankly, I had expected that the spinning disk manufacturers would merge or joint venture themselves out of existence.

That's my feeling as well. None of the traditional large HDD makers have moved portions of their operations to SSDs yet, despite having years of experience with drive firmware and compatiability with ATA/SCSI/SAS/SATA controllers.

Anyway who's left in the traditional HDD market?
Seagate, Hitachi, Samsung, WDC?

Yet with SSDs we have over a dozern manufacturers, including Intel...

[Stereodude]

That's my feeling as well. None of the traditional large HDD makers have moved portions of their operations to SSDs yet, despite having years of experience with drive firmware and compatiability with ATA/SCSI/SAS/SATA controllers.Uh... WD bought Silicon Systems and now sells SSDs. Samsung also makes SSDs.

Yet with SSDs we have over a dozern manufacturers, including Intel...Yeah, but they're all basically selling the same drives. With the exception of Intel and Samsung who are rolling their own controllers, there's only the Indilinx, and SandForce based drives regardless of whose name is on the drives right?

I would think that can't last very long due to competitive pressures. The market will probably thin out to only a few players who can survive. Of course I could be wrong since the RAM market never collapsed to only a few players.

[CougTek]

Anyway who's left in the traditional HDD market?
Seagate, Hitachi, Samsung, WDC?
Toshiba still makes notebook hard drives. And they make SSD controllers has well.

[Chewy509]

Uh... WD bought Silicon Systems and now sells SSDs. Samsung also makes SSDs.


Thanks, didn't know that. I haven't seen any reviews of their drives, so wasn't aware they did them.

[LunarMist]

There are piles of Canvios in some stores.

[MaxBurn]

My take on it is about any electronics house can put together SSD's on a similar assembly line to about any other electronic stuff. But making mechanical disks is a really impressive process with precision machining and clean environments etc. Sort of like a modern day mechanical watch maker.

[ddrueding]

And we've seen how well that worked for the ubiquity of mechanical watches ;)

[MaxBurn]

It's the only type of watch I own but I know it's against the trend. For a while I thought a watchmaker would be a neat retirement job to slip into but these guys have years of training.

[LunarMist]

There are tons of watches with mechancial hand movements around nowadays, although the timing is electronic. Some vehicles have stupid analog clocks too.

[BingBangBop]

Me, I have a dual analog/digital quartz movement watch. I actually have a whole bunch of watches, but that one is my favorite. you can tell it is my favorite from the fact that the plating is wearing off giving it a two-tone color.

However, the buttons on it are all clogged up and no longer work preventing me from setting anything. The watch isn't valuable enough to justify the cost of fixing it. I think those buttons that digital watches use are their biggest flaw! So I just use the analog portion because it uses the standard stem system and that still works.

P.S. Mechanical watchmaking is a dying art. Long term, It is a bad field to get into because electronic movements are so much cheaper. Getting into a field like that, as a retirement job, would be like buying into a slide-rule company...

[MaxBurn]

I recognize this is the wrong forum for this but mechanical watches are never going away. Just try to look at it as a form of art and a mechanical achievement.

Anyway because of these things we are never going to see a new mechanical hard drive maker, they will join and get smaller in volume but I think they will generally die out as we figure out the cost/size issue of SSD.

[Stereodude]

Yeah, but no one looks at their spinning HD as an elegant art form or a mechanical achievement.

Personally, I think spinning disk storage will be with us for quite a while longer, since I don't see flash technology getting cheaper than spinning disks any time soon. A key reason is that flash process technology is already to about the smallest lithography that can be made with visible light and is about to hit the brick wall. Achieving higher densities will require some significant advances. That or they'll move from MLC to QLC or even less reliable flash.

[MaxBurn]

Right so when the mechanical drives die out they are going gone, that's where that similarity breaks down.

[jtr1962]

The biggest difference between mechanical hard disks and mechanical watches is the fact that watches are fairly low-tech. One person and the right tools can grind gears, make springs, pretty much do the whole operation. The raw materials for this have many other uses, and will always be readily available. Any person investing time and a few thousand dollars in tools can make mechanical watches just for the sheer love of it. Incidentally, with clocks being in every electronic gadget these days, watches in general are dying out. I never wear one any more, and I don't even carry a cell phone. It's easy enough to find a clock somewhere, assuming I don't have my GPS on me which I usually do. Truth is the rare times I go out for fun, I couldn't care less about the time anyway.

Mechanical hard disks on the other hand require expensive, sophisticated machinery to make. They also require a clean room. Once they are no longer economically viable ( presumably because SSD hit the same cost per GB ), there will be no incentive for a profit-driven company to make them. And there will be no way an individual who is in love with the technology to make them either ( unless that person is a multimillionaire ). So yes, once they're gone, they're gone. I can even see small runs of incandescent lamps being made for die-hards who refuse to use LED eventually more than I can see hard disks being made. Most people only care that their data is reliably and cost-effectively stored. The technology used is irrelevant.

[CougTek]

...( presumably because SSD hit the same cost per GB )...
I don't see that happening any time soon. And when it'll happen, it will be because mecanical hard drives' cost will increase. We're at 32nm process and the biggest drives are only 256GB. And if you think about it, it takes much more space to create a byte with ROM cells than to create one on a traditional platter. With similar level of technology, mecanical hard drives will always have a space advantage (at least in the foreseable future).

[LunarMist]

Manufacturers can stack many chips in a small space if that is warranted. Cost is the issue.

[sechs]

Agreed.

Spinning disks will stick around as long as they are sufficiently cheap enough. Even if they can't pack the bits much closer together, making flash will continue to get cheaper.

On the other hand, it really is getting harder to pack the bits closer together for spinning disks, and it doesn't get that much cheaper to produce them. It seems to me that we're approaching a plateau in platter density; the nearly annual doubling is gone.

[LunarMist]

The manufacturers have finally agreed on the HAMR.

[jtr1962]

Spinning disks will stick around as long as they are sufficiently cheap enough. Even if they can't pack the bits much closer together, making flash will continue to get cheaper.

On the other hand, it really is getting harder to pack the bits closer together for spinning disks, and it doesn't get that much cheaper to produce them. It seems to me that we're approaching a plateau in platter density; the nearly annual doubling is gone.
That's exactly the problem. None of the technology roadmaps I've seen can get us much past 10 TB drives. And nearly all of them will increase the cost per drive, even though cost per GB will still drop due to increased areal density. Also, we're already experiencing reliability problems from ever increasing platter density. That may well be the limiting factor. Getting something to work in the lab, or even in production, is one thing. Getting it to last for 3 to 5 years in the real world is quite another.

Flash memory on the other hand is becoming less expensive due to economies of scale. Even now, if cost were no object, you could pack enough flash chips into a drive to get well past 10 TB. Nothing is stopping us from putting ten or more stacked flash chips in a single IC package except cost and/or demand. When the flash chips cost ten cents each instead of a dollar, there will be incentive to do so. Undoubtedly we'll reach ever smaller feature sizes because this is needed for other electronic devices, not just flash RAM. Not sure what the inherent limit is here, but it's way smaller than magnetic bits on a spinning disk. Whatever the limits of flash ( or whatever technology may replace it ), the limits on spinning disks are even more daunting. I would be surprised if spinning disks existed much past 2020, with 2015 being a more likely time where they will be phased out. Remember how people talked about CRTs remaining with us for much longer? Even 5 years ago there were people saying CRTs would be with us for a while (http://www.post-gazette.com/pg/05227/552684.stm). And now they're completely gone, indeed have been gone long before the predictions said they would be. I think it's going to be the same for magnetic hard disks.

[LunarMist]

I'm not so sure it will be as soon as many think. Not that much SSD capacity/price progress has been made in the past year. 10 years ago many were proclaiming that tape would be a dead storage medium and it is still progressing.

[sechs]

I dunno. I can buy a 60GB SSD today for about what I paid for a 30GB one a year ago.

I think that what you're really saying is that spinning disks will be relegated to niches. I don't disagree.

[LunarMist]

It is still more than half and meanwhile HDs are cheaper too. At that rate I will be old dead before capacities and price of SSDs and HDs are in parity.

[LunarMist]

3TB (http://www.hitachigst.com/internal-drives/desktop/deskstar/deskstar-7k3000)

[LunarMist]

It is interesting that Hitachi ignored the advancing format this time around.

[Stereodude]

It is interesting that Hitachi ignored the advancing format this time around.I guess they didn't decide to trade error correction information for more user accessible space.

[LunarMist]

Does that mean it is a 5-banger?

[Stereodude]

Isn't that the Hitachi way? 5 platters of love? :batman:

[LunarMist]

Usually, but for the 60GXP.

[LunarMist]

600GB platters at 7200 RPM are still not bad. Do Hitachi drives work in RIAD?

[sechs]

Where's the gigabyte of cache?

[LunarMist]

What GB?

[ddrueding]

You can get my least-favorite internet reseller (http://www.futurepowerpc.com/scripts/product.asp?PRDCODE=1047-0F12450&REFID=FR) to sell you 20 for 1327.80.

[ddrueding]

$681 for 20? (http://www.nextdaypc.com/main/products/details.aspx?PID=A626191&rsmainid=ND0130014)

[Handruin]

That can't be real. That would come out to $34.55 each??

[ddrueding]

Tempted to place the order and take my chances.

[BingBangBop]

That is a price that fits the "too good to be true category". Note they are out of stock which I suspect will never be in stock at least at that price.

[LunarMist]

The normal price is $250 from conducting a few searches.

[Pradeep]

Whilst we wait for that, something available sooner:

"The new Ultrastar SSD400S family delivers the industry’s highest sequential throughput. It is the first to reach up to 535MB/s read and 500MB/s write throughput with 6Gb/s SAS, and 390MB/s read and 340MB/s write throughput with 4Gb/s FC. The new drive also delivers up to 46 000 read and 13 000 sustained write IOPS"

http://www.xbitlabs.com/news/storage/display/20101116191413_Hitachi_Releases_Its_First_Solid_St ate_Drives.html

[Mercutio]

Tempted to place the order and take my chances.

Me too. Hitachi drives are usually great for RAID applications and that would solve the hell out of my disk space problem.

[Handruin]

That is a price that fits the "too good to be true category". Note they are out of stock which I suspect will never be in stock at least at that price.

It also falls into the..."fell off the back of the truck category" (possibly literally).

[Mercutio]

From time to time I have benefited from things that have fallen off the back of a truck. It's been a good, long while, but it has happened.

Anyway, if somebody wants to go halfies on a case of 3TB drives, I'll get in on that.

[Handruin]

Whilst we wait for that, something available sooner:

"The new Ultrastar SSD400S family delivers the industry’s highest sequential throughput. It is the first to reach up to 535MB/s read and 500MB/s write throughput with 6Gb/s SAS, and 390MB/s read and 340MB/s write throughput with 4Gb/s FC. The new drive also delivers up to 46 000 read and 13 000 sustained write IOPS"

http://www.xbitlabs.com/news/storage/display/20101116191413_Hitachi_Releases_Its_First_Solid_St ate_Drives.html


This seems strange to me:


The drives are based on single-level cell (SLC) NAND flash memory produced using 34nm process technology at IM Flash and is powered by controllers developed by Intel.

Why would Intel develop the controller for a competitor that offers a faster product than their own? Does that info give some insight into the next gen Intel SSD drives?

[Handruin]

Anyway, if somebody wants to go halfies on a case of 3TB drives, I'll get in on that.

I would also do the same.

[Mercutio]

Why would Intel develop the controller for a competitor that offers a faster product than their own? Does that info give some insight into the next gen Intel SSD drives?

Maybe the other guys are just faster at getting products into the retail channel, or Intel is practicing strict adherence to a roadmap in defiance of engineering or manufacturing realities?

[ddrueding]

IIRC, Intel got into the SSD business because no one was stepping up with good drives.

[LunarMist]

Whilst we wait for that, something available sooner:

"The new Ultrastar SSD400S family delivers the industry’s highest sequential throughput. It is the first to reach up to 535MB/s read and 500MB/s write throughput with 6Gb/s SAS, and 390MB/s read and 340MB/s write throughput with 4Gb/s FC. The new drive also delivers up to 46 000 read and 13 000 sustained write IOPS"

http://www.xbitlabs.com/news/storage/display/20101116191413_Hitachi_Releases_Its_First_Solid_St ate_Drives.html

I posted about that in the SSD thread. :)

[LunarMist]

I would also do the same.

I don't know how I could attach them to my computer, but a few would be nice. :compress:

[Stereodude]

I don't know how I could attach them to my computer, but a few would be nice. :compress:Duct tape or velcro would probably do the job. ;)

[LunarMist]

I was thinking perhaps some type of bridge broad (e.g., USB 3.0) that circumvents the evil 2TB drive limit would be nice.

[mubs]

I was thinking perhaps some type of bridge broad (e.g., USB 3.0) that circumvents the evil 2TB drive limit would be nice.
Now you're talking dirty, Lunar.

[LunarMist]

Maybe I'm not being clear or not understanding. Is there something that can be done in the PHY instead or with a specific combination?

[mubs]

I was thinking perhaps some type of bridge broad (e.g., USB 3.0) that circumvents the evil 2TB drive limit would be nice.
:)

[LunarMist]

Even with the typo, how is that dirty? Maybe mildy derogatory, but not more.

[mubs]

Just messing with you, Lunar, sorry about that. I'll shut up now.

[BingBangBop]

$681 for 20? (http://www.nextdaypc.com/main/products/details.aspx?PID=A626191&rsmainid=ND0130014)

I emailed them about it. That is the per drive price when you order at least 20. i.e. bad bargin.

[ddrueding]

Indeed. Thanks for the follow-up.

[Handruin]

Oh my! I was expecting them to be around the $299 range, not $681/per drive. Thanks for emailing them and finding out the scoop.

[LunarMist]

http://eshop.macsales.com/item/Hitachi/0F12450/

http://www.allstarshop.com/shop/Hitachi-0F12450_p29542_afg.html

[Mercutio]

Goddammitsofuckingmuch (http://www.wdc.com/en/company/pressroom/releases.aspx?release=ba433e4b-bff8-4d99-b60f-7f02aa42f444)

[Handruin]

Wow, I wasn't expecting that one.

[Will Rickards]

So we have Samsung, WD, and Seagate to choose from now? At least in the spinning disk area. Which at this point must be pretty low margin.

[Handruin]

I think there is still Toshiba and Fujitsu but I don't know many who still use them.

[Mercutio]

There is only Samsung now.

[time]

An odd couple indeed. This is a marriage between a marketing machine and an engineering shop, with the marketers calling all the shots. Just looking up specs highlights how opposite are the corporate cultures: Hitachi (née IBM) publishes just about every parameter you could wish for, while with WD you can't even find out how fast the platters spin, and a model number alone may not be enough to identify a specific model. :(

The fact that HitachiGST product releases in the last couple of years have been increasingly behind the curve suggests they've been saving pennies in anticipation. Apart from enterprise cred, can anyone work out what WD gets out of this? It's hard to see any of the Deskstar or Travelstar models continuing; even the famous 5-platter version isn't necessary when your areal density is high enough. :(

[Will Rickards]

Maybe this is a good thing and new WD drives won't suck?

I think WD gets bigger as a player to compete better with Seagate. I hope this will increase their design capability and manufacturing quality.

[Stereodude]

I think there is still Toshiba and Fujitsu but I don't know many who still use them.Those two merged also, so it's really just Toshiba.

[Bozo]

IBM (Hitachi) and WD have been in bed with each (http://www-03.ibm.com/press/us/en/pressrelease/2684.wss) other for a long time.
WD and Hitachi both had 5 pin molex connectors on their SATA drives for a while. Along with a SATA power connection.

[LunarMist]

F******************ck!

[sechs]

There goes the neighborhood.

When is that next gen of SSDs due out? You know, the ones not by the spinning disk manufacturers....

[LunarMist]

Maybe this is a good thing and new WD drives won't suck?

I think WD gets bigger as a player to compete better with Seagate. I hope this will increase their design capability and manufacturing quality.

Seagate sucks anyways, and Samsung is always behind the technology curve. Elimination of Hibachi will increase prices and slow down hard drive improvements due to lack of competition. I wonder if WD will retain the 5-platter mechanism. :crucified:

[LunarMist]

I'm thinking about buying 5 of these (http://www.newegg.com/Product/Product.aspx?Item=N82E16822145493). Is it a good idea or not?

[ddrueding]

Certainly a great price. 3TB drives do still have issues with some controllers, and all RAID controllers. IIRC, you don't use RAID, and you use newer Intel chipset boards, so you should be fine.

[LunarMist]

Certainly a great price. 3TB drives do still have issues with some controllers, and all RAID controllers. IIRC, you don't use RAID, and you use newer Intel chipset boards, so you should be fine.

I would be using RAID controllers and some overlays I guess, but no RAID modes. Does USB 3.0 work with those drives?

[ddrueding]

I've run 3TB drives on USB without issue, though not those exact drives. I have yet to hear of any RAID controllers that correctly identify and work with 3TB drives of any kind. I would be very cautious in that aspect.

[Mercutio]

I don't think the 5400rpm drives are reliable enough for general use. I think they're made from manufacturing rejects or something, based on the staggering failure rates I've had with them across Seagate, Hitachi and Samsung models. I should not have to RMA 20% of any model of drive I buy, but all of them are at that level.

[CityK]

Mercutio;So we have Samsung, WD, and Seagate to choose from now? At least in the spinning disk area. Which at this point must be pretty low margin.


There is only Samsung now.

And then, soon maybe none (http://67.222.3.121/%7Ehandruin/forum/showthread.php?p=146089#post146089)

[LunarMist]

Holy moly! So there will be only one global hard drive monopoly? Meanwhile flash memory durability is crapping out at smaller nodes. Basically all of our data is screwed. :cry:

[Stereodude]

Holy moly! So there will be only one global hard drive monopoly? Meanwhile flash memory durability is crapping out at smaller nodes. Basically all of our data is screwed. :cry:By my count that would leave us with two. WD and Seagate.

[Chewy509]

By my count that would leave us with two. WD and Seagate.

Do Toshiba still make drives? (IIRC, Toshiba purchased Fujistsu storage)...

[Stereodude]

Do Toshiba still make drives? (IIRC, Toshiba purchased Fujistsu storage)...Make that three then. ;-)

[LunarMist]

OK, somehow I am confused about Seagate absorption into WD. Anyway, Toshiba/Fujitsu are not in the 3.5' drives.

[Chewy509]

OK, somehow I am confused about Seagate absorption into WD. Anyway, Toshiba/Fujitsu are not in the 3.5' drives.

I thought they were?

http://www.toshiba.co.jp/about/press/2010_12/pr1401.htm

Seems to indicate 2TB 7200rpm drives?

[LunarMist]

Hmmm. I wonder why they are not sold in the US, or maybe the 3.5" drives are new.

[LunarMist]

Hibachi backs up in the clouds? Sounds like BS to me.