backtop


Print 41 comment(s) - last by Zaphod Beebleb.. on Jun 26 at 1:36 PM


HRD Schematic  (Source: DataSlide)
HRD is capable of 160,000 IOPS according to developer

For many years now the main storage method used for computers have been hard drives that use spinning magnetic platters as the storage medium. Over the last few years, the solid-state drive has come to market with much faster performance and a much higher price tag.

A company called DataSlide has unveiled a prototype of a new storage device that may one day replace both HDDs and SSDs. The prototype technology is called Hard Rectangular Disk or HRD. DataSlide says that the technology is patented and can achieve 160,000 IOPS and 500MB/sec performance levels while consuming under 4 watts of power.

The device consists of a piezoelectric actuator that keeps the rectangular media in precise motion, a diamond solid lubricant to protect the surfaces (which are in direct contact) and a massively parallel 2D array of magnetic heads for reads and writes to up to 64 embedded heads at a time.

ZDNET reports that DataSlide is also working to incorporate the technology into a "smart" storage unit for use in I/O intensive multiple concurrent stream applications. The new device is at the prototype stage, but uses currently available production technology. The technologies used in the HRD include perpendicular recording media, semiconductor lithographic heads, and LCD glass treatments.

The concept for the HRD was taken from IBM's Millipede concept and then reworked with common and available technologies. The prototype has a long way to go before it reaches a final production stage and the company gives no idea as to how long it might take before the HRD comes to market. ZDNET reports that the HRD has a capacity of 36GB in its prototype stage.



Comments     Threshold


This article is over a month old, voting and posting comments is disabled

Huh? What? Why?
By Belard on 6/18/2009 4:55:28 PM , Rating: 3
Have a large surface area move while the 64 heads are stationary? There would have to be quite a bit of open area to move this "card" platter in both directions.

And with 64 heads, if one fails a chunk of the drive is unreadable. The motor/magnets to move it can fail.

The HD is reliable (mostly) because its a simple design. Its not much different from a record player. Spin discs in a circle and slightly move the heads.

SDD is still the future, no moving parts. Sure its expensive... today. But it'll be cheap before this square drive ever make it to market.




RE: Huh? What? Why?
By ClownPuncher on 6/18/2009 5:02:22 PM , Rating: 5
We Americans already have too few moving parts, which is why diabetes and heart disease are such concerns.


RE: Huh? What? Why?
By masaauk on 6/18/2009 5:39:18 PM , Rating: 2
Why would you vote him down? I thought what he said was hilarious... does nobody understand sarcasm anymore?


RE: Huh? What? Why?
By feraltoad on 6/19/2009 1:34:44 AM , Rating: 5
That wasn't sarcasm. Wait, is this sarcasm? Oh, your good.


RE: Huh? What? Why?
By xsilver on 6/18/2009 11:22:18 PM , Rating: 2
Apparently bowel movements are still up 3% from last year :P


RE: Huh? What? Why?
By MrPoletski on 6/19/2009 5:01:30 AM , Rating: 5
87.7% of all statistics are made up on the spot but close to 100% of all turds are made up on the squat.


RE: Huh? What? Why?
By Belard on 6/19/2009 7:13:50 AM , Rating: 1
Wish I could vote you up for that one. :)


RE: Huh? What? Why?
By UNHchabo on 6/18/2009 5:08:34 PM , Rating: 3
I think any technology is worth chasing if it's possible that it will yield better performance. Even if it is more unreliable and costs more, there will be buyers who just care about performance.

Because of reliability concerns (which you brought up) inherent with moving parts, I can't see this replacing regular HDDs and SSDs in the general use, but you never know...

Disclaimer: I work for Intel, but this post is based on my own opinions and observations. I haven't even had a chance to use an SSD yet! :)


RE: Huh? What? Why?
By futrtrubl on 6/18/2009 7:11:38 PM , Rating: 2
Except that this will be less reliable, (may) cost more and have just about the same performance, if not less by the time it comes out. It's set to be "years" to retail so by that time SSDs (currently threatening to saturate 300MB/s SATA 2) should have passed it in speed. If this has a place when it comes out I would think it would be as a cheaper/higher capacity alternative to SSDs which would indeed make it a replacement for HDDs or as something in between.


RE: Huh? What? Why?
By retrospooty on 6/19/2009 12:32:43 AM , Rating: 5
"Except that this will be less reliable, (may) cost more and have just about the same performance, if not less by the time it comes out"

Says who? It's not even out yet, how could you possible know it's reliability ratings, much less speed and cost?


RE: Huh? What? Why?
By afkrotch on 6/19/2009 3:33:58 AM , Rating: 2
Says math. More read/write heads than regular hard drive. Higher failure rate.

Compared to an SSD, I don't know.


RE: Huh? What? Why?
By tastyratz on 6/19/2009 8:06:52 AM , Rating: 1
Math AND logic

Direct contact with the platter? I'm sorry but unless its user serviceable no level of lubricant will take the direct contact of all those write heads without eventually wearing a trench in the platter.
On that, who would want to change their oil "every 3000 iops"? If they could get these to have reasonable longevity I would be extremely surprised.

This is something that will likely be too little too late.
What has always surprised me is the fact that the spinning platter design only has 1 armeture of heads. If they took the recent increases in data density then translated it to smaller diameter platters instead of less platters they could maybe squeeze an extra head assembly on the other side. This would dramatically increase iops.


RE: Huh? What? Why?
By MadAd on 6/22/2009 6:36:29 AM , Rating: 2
or even 2 sets of heads ive always thought, one arm on the other side, 2x the data transfer no?


RE: Huh? What? Why?
By Oregonian2 on 6/18/2009 10:17:32 PM , Rating: 2
Somehow this product, to me, has the "feel" that the RCA video disc had (basically a high bandwidth version of an LP audio disk, complete with mechanical needle).

My WAG is that it will likely be as successful. TBD though.


RE: Huh? What? Why?
By inperfectdarkness on 6/18/2009 11:44:58 PM , Rating: 2
chasing this is like rooting for RDRAM.


RE: Huh? What? Why?
By mmcdonalataocdotgov on 6/19/2009 7:16:53 AM , Rating: 2
Are you Italian by any chance? This reminds me of the Alfa Romeo philosophy: "The most exhilerating automotive performance achievable, for the shortest period of time possible." - Jeremy Clarkson.

Another answer to a question nobody asked, like the Bob interface or Bernoulli drives.


RE: Huh? What? Why?
By Belard on 6/19/2009 8:40:09 AM , Rating: 5
I studied the drawing quite a bit more in detail and later read the posts below. My opinion changed before I read the responses, but having seen how tech gets developed and dropped - failure is the norm.

The concept is valid. The drive would have "thousands" or so heads and the movement is slight. Had this drive was at this phase about 7~10 years ago and came out before 2003, I'd say it would have a chance.

- Its still a mechanical device.
- Its still made of many parts.
- While it should NEVER have a "head crash", the heads themselves can fail.

The problem is still TIMING, that is always the key. The netbooks came out at the right time with the right tech to make excellent market share.

- SSD or other non-mechanical storage devices are the future.
- SATA 3/6 Gbit is out in the next year and some SSDs will easily hit that.

- As someone said about failure of an HD and recovery from the platters. But many HD failures is having heads crashing which causes loss of data. And that it has moving parts and subject to vibrations can cause failures.

SSD, by having NO moving parts, NO air-holes makes it a very reliable device... other than limited writes. When SSDs can prove to survive daily writes 24/7 for a set number of years in which desktop life is 10+ years will be an important level for SSD.

Also, a 500GB SSD unit needs to cost under $200, or at least 120GB at $75. We need to remember that notebooks can make use of these drives first since 2.5" drives are still "slow" compared to desktops. So a 120GB at a low price is important. Remember, about 1 year ago a 120GB notebook drive was about $100. Current prices of 128GB SSDs is $250~400 so we're about 12 months away from hitting the $100 mark.

A good transition setup is to have a 32~64GB when they hit $50~60 and then use a standard 1TB for data. That'll make Vista/Win7 fly while keeping the costs down and when the SSD "wears out" in 2-3 years the drives should be cheaper.

Lets look how tech is different from 2000 compared to what we have today. We went from CD-Rs to DVD-Rs to Flash Drives. Music CDs to MP3s players. Mono-colored phone that mostly just DIALED or Re-dialed to GPS/Game/Camera/Internet browsing phones. 20GB HDs to 2000GB HDs. 3D Video cards that are hundreds of times faster than the Voodoo2.

By 2020, who knows what we'll have. I think a POWER Users computer will fit in a keyboard with a 4TB SSD drive that are 1.8" in size and the whole "PC" will cost under $1000. No blu-ray drive sucking up space. Just memory cards.


RE: Huh? What? Why?
By lewisc on 6/21/2009 2:44:03 PM , Rating: 2
...and that fabled, keyboard-shaped computing powerhouse shall be called the Commodore 64.


RE: Huh? What? Why?
By StoveMeister on 6/18/2009 5:15:23 PM , Rating: 5
quote:
There would have to be quite a bit of open area

Nope- uses a piezo-electric actuator- motion of the centre platter would be measured in microns at most. Its not moving the disk around to align under any single head- it has a MASSIVE amount of read/write heads on the read plates. the "64 at once" thing would be just a limitation of the way the heads are addressed.
quote:
And with 64 heads, if one fails a chunk of the drive is unreadable. The motor/magnets to move it can fail

Again- nope. For starters there are WAY more than 64 heads- only 64 can be addressed simultaneously. If one head fails, only the area of disc below that head is unreadable- and that assumes worst case scenario. If the disk moves far enough the area may traverse 2 or 3 read heads, so all a head failure would mean is a slightly greater lag in one area of disk. Not too dissimilar to a bad sector on a HDD I'd say.
Basically the main area I'd be worried about is the lube- contact means wear.


RE: Huh? What? Why?
By Souka on 6/18/2009 5:37:42 PM , Rating: 4
Great... I can see it now.. yearly oil-change for your HD.. heh.

;)


RE: Huh? What? Why?
By TennesseeTony on 6/18/2009 5:38:33 PM , Rating: 2
Diamond to Diamond contact does not mean abrasion, if it's smooth enough. Probably some sort of super thin, two atoms thick, perfectly smooth plating process.


RE: Huh? What? Why?
By icanhascpu on 6/18/2009 8:11:14 PM , Rating: 2
Nothing is perfectly smooth, but I know what you mean.

I really dont think this is going anywhere. SSD IOPS will go up with throughput while price comes down. I dont see this coming out in time to have any real advantages.


RE: Huh? What? Why?
By gmyx on 6/19/2009 8:45:09 AM , Rating: 2
quote:
Nope- uses a piezo-electric actuator- motion of the centre platter would be measured in microns at most. Its not moving the disk around to align under any single head- it has a MASSIVE amount of read/write heads on the read plates. the "64 at once" thing would be just a limitation of the way the heads are addressed.


Exactly. According to this slide: http://www.dataslide.com/images/Oracle-Embedded-Be... the disk moves by 250 micrometers (a.k.a microns), enough to cover an entire sector.

quote:
Again- nope. For starters there are WAY more than 64 heads- only 64 can be addressed simultaneously. If one head fails, only the area of disc below that head is unreadable- and that assumes worst case scenario. If the disk moves far enough the area may traverse 2 or 3 read heads, so all a head failure would mean is a slightly greater lag in one area of disk. Not too dissimilar to a bad sector on a HDD I'd say. Basically the main area I'd be worried about is the lube- contact means wear.


The disk only travels enough for one sector to be read (one head per sector). Just like HDDs, you need some form of ECC to ensure that your data is not lost by a single dead head. Since they are targeting Oracle DBs, I'm betting that a portion of the device is dedicated to ECC.


RE: Huh? What? Why?
By Jimbo1234 on 6/20/2009 3:34:15 PM , Rating: 2
quote:
Exactly. According to this slide: http://www.dataslide.com/images/Oracle-Embedded-Be... the disk moves by 250 micrometers (a.k.a microns), enough to cover an entire sector.


250um (microns) is 1/4mm. That's a lot of motion for a piezo actuator, unless it's a "walking" type actuator, which is expensive. I just purchased 6 (with 3 controllers) for work for a total of just under $50,000. Granted these are UHV and bakable actuators, but the cheap versions cost about $1500 each. If this walking piezo is based on the two mode resonance model, then reliability will be its Achilles.

A piezo stack actuator would require 1mm of length per 1um travel, making this thing huge. And piezo stacks are nonlinear, adding more headaches.

It's nice to see alternative technologies, but I do not see this being very promising mostly due to the issues with piezo actuators.


RE: Huh? What? Why?
By Belard on 6/19/2009 8:47:21 AM , Rating: 2
I've responded to someone else but my response to you is the same... as I kind of responded to both of you. :)

I did read what you said and had looked at the image. Yeah, I think its an interesting good idea. But its late.

Look for the post by the intel guy.

The developers of this square HD don't know when it'll be market-ready. That means 2~5 years to me. SSDs at 120GB are at $250 for low end but in two years a 256GB SSD should cost under $100 easily and hitting SATA 3 speeds.


RE: Huh? What? Why?
By afkrotch on 6/19/2009 9:15:55 AM , Rating: 3
The problem is you don't know what will happen. What if this drive comes out at 2-3 TB and with 500 MB/s transfer speeds?

It's really all just guess work, but I won't condemn it. Anything that's faster or larger is always a plus. If SSDs came in the 500+ GBs with a reasonable price, I'd get one. As of right now, they just aren't attractive to me.


RE: Huh? What? Why?
By Belard on 6/20/2009 7:15:28 AM , Rating: 2
The issue with MOST current SSDs is that the tech is new and its limited by SATA 2/3Gbs interface. Found out the SATA Gods says that "SATA II" or (2) is incorrect and that SATA 3 should be called SATA 6GB. Oh heres a paste:

"The specification should be called Serial ATA International Organization: Serial ATA Revision 3.0, and the technology itself is to be referred to as SATA 6 Gbit/s. A product using this standard should be called the SATA 6 Gbit/s [product name]. The terms SATA III or SATA 3.0, which are considered to cause confusion among consumers, must not be used.[" Hows that for anal?

Okay, Revision 3 allows up to 6GBit/s transfers. So whats the confusion?

Anyways, high end SSD are easily faster than 500MB/s. Some of the top consumer SSD-PCIe units perform about 1,500MB/sec, far beyond the abilities of SATA 2 or 3. Size range 120GB~1TB+. http://www.engadget.com/2009/03/11/fusion-io-break...

Of course these "drives" costs thousands of dollars. But eventually we'll get there. Remember, 4 years ago: 512mb Flash drives were $50~75. $10 buys you 4GB. 10 years ago, $250 bought you a 6GB HD. Today $50 gets you 500GB.

SSDs have been advancing properly and will accelerate as more companies join. Look at the companies we suddenly have in the storage market!

To have a 500GB SSD thats $100 and transfer data at 500MB/s would be sweet (more than double than intel's top drive), and having a $150 version than does 1000+MB/s would be worth it. Boot Vista/Win7 in about 5 seconds. Scanning for viruses would be as fast as the CPU can keep up, etc.

We're still 2 years away.


RE: Huh? What? Why?
By MatthiasF on 6/18/2009 8:10:43 PM , Rating: 2
You complain that this new type of drive might cause problems if a head fails, but what happens if a cell on your SSD fails?

At least with magnetic discs and even this new type, you can remove the media for recovery. That's not an option with SSD.


Is this really a new design ?
By greylica on 6/18/2009 8:19:29 PM , Rating: 2
Well, the only difference I see in this approach, is that the heads are moving instead of the magnetic plate itself. If we could use diamond coated plates inside hard drives, many more heads per drive, using level triggered bits,(diferent times between heads read and write bits), wouldn´t be the same thing we use today called hard drives with a boost ?.

What´s the difference between a head moving and a plate moving ?

Sorry, I could not find the difference...




RE: Is this really a new design ?
By greylica on 6/18/2009 8:32:19 PM , Rating: 2
Better than this...
Is there a way to litograph an array of heads for a single plate and stand them together forming a piece with NO ONE moving parts inside ?

This kind of storage will solve the question forever, no moving parts, except the magnetic orientation itself.
At an reasonable space between heads, no data will be lost and we could achieve super fast read/write rates.


RE: Is this really a new design ?
By codeThug on 6/19/2009 12:54:25 AM , Rating: 2
You know..

Cats and Dogs living together. What's the problem.


RE: Is this really a new design ?
By codeThug on 6/19/2009 1:03:23 AM , Rating: 1
Actually I'm getting tired of all this spinning nonsense. It all reminds me of a Wankle engine.

We need a reciprocating style hard drive with pistons and crankshafts and bits slamming into each other at insane speeds. If your database is slow, just shoot a little nitrous into the HDA.

DataSlide... uh.. uh.., God that name just makes me laugh, needs to rethink it's approach.


Product Naming
By SuckRaven on 6/19/2009 5:49:57 AM , Rating: 2
Now, even your computer can have something to get excited about. Everytime you boot up, your computer will get a HRD-ON. LMAO...




RE: Product Naming
By Zaphod Beeblebrox on 6/26/2009 1:36:40 PM , Rating: 2
Software. TeeHee.


Too late.
By icanhascpu on 6/18/2009 7:51:51 PM , Rating: 2
About ten years too late really.

This tech would really have to give us something that SSD doesnt, because SSD is only sitting still where it should be, and thats not in advancements. SSD will already have huge market penetration and very low price by the time this is ready for anyone here. I highly doubt it would even have the $ per GB advantage in 5 years.




By xeroshadow on 6/18/2009 8:40:44 PM , Rating: 2
What happened to IBM's next generation HDD technology they dubbed 'racetrack'? Is this other technology more likely to see production first?




IBM's tech...
By CZroe on 6/19/2009 3:22:37 AM , Rating: 2
The first thing I thought of was IBM's tech where tiny bumps were created and read with tiny "fingers" that could make one sweep and conceivably read the contents of the entire drive in one pass. After all, they sold their HDD division to Hitachi not too long after announcing it. Was that "Millipede?"

If so, this adaptation removes the non-magnetic advantage of IBM's tech and reintroduces that susceptibility. :(




Why on Earth?
By Jackattak on 6/18/09, Rating: -1
RE: Why on Earth?
By bobsmith1492 on 6/18/2009 5:38:12 PM , Rating: 2
It's not really mechanical; sure, there is movement, but it's not spinning; it's more like a MEM device.


RE: Why on Earth?
By CZroe on 6/19/2009 3:26:11 AM , Rating: 2
"It's not really mechanical; sure, there is movement, but it's not spinning; it's more like a MEM device."

MEMS devices are mechanical. MEMS = MicroelectroMECHANICAL systems. I'm not sure why you seem to think that such movement doesn't equate with "mechanics" because it usually does.


RE: Why on Earth?
By Jackattak on 6/18/09, Rating: -1
"Paying an extra $500 for a computer in this environment -- same piece of hardware -- paying $500 more to get a logo on it? I think that's a more challenging proposition for the average person than it used to be." -- Steve Ballmer

Latest Headlines










botimage
Copyright 2012 DailyTech LLC. - RSS Feed | Advertise | About Us | Ethics | FAQ | Terms, Conditions & Privacy Information | Kristopher Kubicki