Multi-head hard drives?

D

Dario D.

Gawd
Joined
Dec 8, 2004
Messages
582
To anyone who really understands hard drives, would it make sense to have multiple head arrays (the reading-things below)? And, specifically, why or why not?

I would imagine that a multi-head hard drive would simply act like RAID, because you'd have multiple heads pulling data off the platters simultaneously.

I've long, long wondered why they don't just do that to multiply the speed of HD's. (realizing also, of course, that they seem to have no intention of allowing hard drives to progress any faster each year than the fixed rate at which they do)

095-hard-drive-maintenance-08.jpg
 
Last edited:
Performance-wise you can get pretty much the same thing using RAID, and adding another set of read/write heads increases complexity and decreases reliability. The result would be a drive that for the most part acted like a small RAID-0, but with a much higher cost per GB and a higher chance of failure.

IIRC this has been tried, but the benefits weren't worth the additional cost. Desktops and servers can easily just use RAID. For laptops, in the past using multiple heads would have resulted in a storage capacity that was simply too small, since they would have to use smaller platters in order to cram a second set of heads into a 2.5" package. There might have been a window recently where such a drive would have been useful (though expensive and less reliable than a normal drive), but SSDs have now closed it.
 
You would technically double your failure rate at this approach
 
The read/write heads do not move independently in current hard drives, and, changing that would be rather difficult. If it were an easy task, it would have been done quite some time ago.
 
accualy i think this was done
on drives 20 or so years ago
drives were considerably longer and complex
company was bought by seagate and tehnology discontinued due considerable increase of platter density
 
There was an article on Tom's that addressed the idea of a dual actuator/head drive. Basically, it's not a good idea because of much added complexity and cost for little benefit.

Hard drives are essentially commodity items anyway; a dual-head hard drive won't sell well simply because it costs more. In the desktop market anyway; in the server market the added complexity and being relatively 'new' will make people think they're less reliable (regardless if it's true or not). Hard drive makers will need to spend a lot of marketing dollars disputing this in order to sell the product.
 
I was thinking the OP was referring to staggered heads on the same armature, each reading offset tracks. That way you could read 2 or 3 parallel tracks simultaneously, and with the large buffers on today's drives, the drive firmware could then reassemble these bits into their proper order. I can see how this could increase sequential performance a good deal, but it obviously won't help random performance at all. Still, it seems it could be worthwhile - there must be a technical reason why this doesn't work.
 
According to that article that SockMan posted (thanks), it all sounds doable (I mean, they've done it), and Seagate never mentioned reliability as a con.

As mirrored by a few commenters in that article, I don't like their answer about why they stopped making those hard drives (sounds like a whitewash).

I really think they just wanted to let their hard drive speeds continue to incrementally increase each year by the same fixed margin that they've ALWAYS increased... like the HDD version of Moore's Law (which is where CPU's and video cards always increase by precisely 50% speed each year, never allowing any sudden technological advancements to boost this curve for any reason). I also find it the most ironic thing in the world that the speeds of SSD's - a COMPLETELY new line of devices - have fallen right in line with current hard drives. (how sad that the infamously slow hard drive STILL won't be breaking free of its speed weaknesses)

Not necessarily in relation to the above paragraph, one commenter pointed out part of what I was thinking:

I still think seagate dosn't have a good answer to this. From a engineering standpoint it is very doable. The technical drawings of the original patent show 2 actuators/heads reading from the same platter. This is not needed in a current HDD, with 500GB per platter technology availible you could, in theory, have a 2 platter design capable of 1TB total storage and have a independent actuator/head for each platter. Like others have said this would be similar to RAID 0 in performance.
Click to expand...

And now, from a logical, non-technical-details standpoint, let's break down what Seagate said:

First and most importantly, it's cost prohibitive to produce such drives. When you factor in the extra components, increased material usage, and weight gains, the advantages are minimal said Seagate.
Click to expand...
Ummm... but it's like building 50% more hard drive, except that you can sell it for more than 2x the price. If anything, by NOT just selling someone multiple hard drives for RAID, you're saving materials, and weighing LESS (because it's just 1 drive). - And developing this isn't like the task of developing SSD's... they already got it working.

Secondly, dual-actuator drives will be larger, produce more heat, and consume more power
Click to expand...
2 things... First of all, say that to my GTX 260 video card, and second, a multi-head drive will NOT be larger than multiple drives in RAID getting the same speed, NOT produce more heat because of that, and NOT consume more power. (regarding size, they could even just make the platters smaller, like in notebook hard drives) If anything, a multi-head drive would be the eco-friendly, cost, energy, and materials-efficient alternative to multiple hard drives running in RAID. (or just become the gamers' standard "performance" hard drive, like the "performance" anything else in a gaming computer)

I know that my limited, non-engineer perspective doesn't cover all the bases, so I would really like for someone who *actually* knows the internals of these things to elaborate on whether or not Seagate is pulling a fast one. I really think hard drive manufacturers don't want the existence of fast hard drives that break their current trend of doing fixed-paced baby-steps each year. (like the CPU and GPU makers, who live by Moore's Limiter)
 
Last edited:
Nowadays, SSDs make it more unlikely that any more research will go toward multi-head drives. That's where the enthusiast and ultra-high-end performance crowds are now. The more conservative crowd - businesses and low/mainstream users - will stick to proven hard drive technology where hard drives are mostly commodities anyway. Seagate will probably put more effort to sell to businesses anyway - a market where there's no room for 'unproven' and bleeding edge improvements to old technology.

Seagate probably can make a dual actuator/head drive nowadays, but they won't for purely business reasons.

Research on improving conventional hard drives is fairly stagnant - the last big innovation seemed to be perpendicular recording, but even that was just to keep hard drive capacities from hitting a brick wall. There's only so much you can improve a technology before you need something completely revolutionary to replace it - this applies to any technology, not just hard drives. There's a lot more room for innovation early on when the technology is new and exciting. However, Innovation and revolutionary advances are made less frequent as technologies mature and mutli-billion-dollar business arise that depend on the sale of proven technology.

In the case of magnetic media, perhaps research into Hard Rectangular Drives will be more worthwhile?
 
Nowadays, SSDs make it more unlikely that any more research will go toward multi-head drives.
Click to expand...
Yeah... I was *completely* stoked about the arrival of SSD's (I've been calling HDD's "record players" for years. That's exactly what they are), but was sooooooooo disappointed to see their speeds fall right in line with existing hard drives. So, now I just hope that someone comes along with an SSD that forces all the other companies to drop their "inching forward" practices, so we can get some speed in our stinking "hard drives".
 
Dario D. said:
Yeah... I was *completely* stoked about the arrival of SSD's (I've been calling HDD's "record players" for years. That's exactly what they are), but was sooooooooo disappointed to see their speeds fall right in line with existing hard drives. So, now I just hope that someone comes along with an SSD that forces all the other companies to drop their "inching forward" practices, so we can get some speed in our stinking "hard drives".
Click to expand...

But people don't buy SSDs for sequential transfer rates, they buy it for random access speeds. That's where SSDs completely blow away hard drives and is the one area where traditional hard drives can never compete. No amount of incremental improvements over traditional spinning disks will ever improve access times to SSD-levels.
 
A big reason why SSD falls back to HDD levels is simply due to SATA. No one imagined (when it was drafted) that SSD would totally obliterate what SATA was able to provide. SATA's rate of progression was based on the assumption that it would be fed by HDDs. SATA2 was and still is plenty for the traditional drives it was meant for.
 
Dario D. said:
Yeah... I was *completely* stoked about the arrival of SSD's (I've been calling HDD's "record players" for years. That's exactly what they are), but was sooooooooo disappointed to see their speeds fall right in line with existing hard drives. So, now I just hope that someone comes along with an SSD that forces all the other companies to drop their "inching forward" practices, so we can get some speed in our stinking "hard drives".
Click to expand...

I don't see much effort being put towards developing HDDs that push the performance envelope now that SSDs are taking over. At least in the consumer market. People are fascinated with the speed, silence, and reliability of SSDs. Therefore, as I said in another thread, I doubt the engineers at WD or Seagate are working late nights trying to develop a 20k RPM consumer drive.

If you want more speed out of platter based drives, why not go for a 15k SAS setup? Or are you wishing for hard drive innovation to provide SSD-like performance from inexpensive platter-based drives? I'm just not sure if it's practical for HDD manufacturers to try to go in that direction when SSDs provide so many advantages. Should they devote resources trying to improve old, yet established technology when the market is clearly going in another direction? Like you said, recent advancements in consumer HDD tech haven't been that substantial other than perpendicular recording and the 10k RPM Raptors. Meanwhile, in the relatively short time that SSDs have started to become a viable solution the industry has been advancing by leaps and bounds. If WD or Seagate managed to release a HDD with some new tech that managed to give the good SSDs a run for their money, I'd be willing to bet it wouldn't be on top for long because SSDs are advancing at a much faster rate. So again, I'm not sure it would be in their best interests to do something like that. Don't get me wrong, platter-based drives still sell very well but that's due largely in part to the fact that most average Joes aren't willing to pay more for less space, even if it is faster. However, more and more people are becoming interested in SSDs and I think a LOT of people will be making that transition during the next 12-24 months.

The focus right now is on making SSDs bigger, better, and cheaper so that more people will opt for them vs. mechanical drives. I know I'm wishing for bigger/better/cheaper SSDs a lot harder than I'm wishing for someone to develop a platter-based drive that competes with them. SSDs are already more than capable of saturating SATA II...sounds like they're advancing so fast that other tech is going to have to end up catching up with THEM vs. the other way around as was the case just a few months ago.
 
I don't see much effort being put towards developing HDDs that push the performance envelope now that SSDs are taking over. At least in the consumer market. People are fascinated with the speed, silence, and reliability of SSDs. Therefore, as I said in another thread, I doubt the engineers at WD or Seagate are working late nights trying to develop a 20k RPM consumer drive.
Click to expand...
Yeah, of THAT I'm certain. I could still see a line of "performance" multi-head hard drives coming out for gamers and designers (people who work with mass data and HD video, like me, and need MASSIVE read/write speeds) but I don't see most companies so easily wanting to distract from SSD's (although, currently, I just don't have enough incentive to BUY an SSD - and WON'T until they reach at least 1TB, and for a non-gouging price - so my thought is that they might as well push for regular, rotating Performance drives, since the mainstream gamer/designer/pro (or average user for that matter) isn't going to get serious about SSD's anytime soon, on top of the growing apathy toward today's "standard" hard drives).

I've been holding back on upgrading my hard drives for almost 2 years now, just because of the state of stagnation the hard drive industry is sitting in. My 2-year-old drives are BOTH still some of the fastest/largest there are. From my perspective, SSD's won't even be competing with the space/price of standard drives for YEARS still (thanks to the hard drive version of Moore's Law), and I can't just buy an SSD that *only* offers speed, if it doesn't have massive space and at least half-decent price to go with it. It's out of the question.

Hard drives are the achilles heel of computers, partly for reasons that I personally think shouldn't even be in the picture. I think the drastic limits of today's SSD's (space/price) are limiting these companies more than anything... The average computer user isn't excited about SSD's, sees NO real reason to buy one (in fact, the price + lame storage space repels people), and so if anything, the HDD makers are killing the public perception of the "brand" of SSD's - something that could be devastating, if they don't do something quick to give them some zazz. (in fact, if I were an HDD maker, I wouldn't even call them "SSD's" - what kind of retarded, cryptic phrase is "Solid State Drive"?... I'd call them Ray Drives or something, and bundle them in video-card style boxes, and of COURSE make them powerful enough to keep their buzz alive longer than a day. I mean, word about the these things (speed/space/price that all MEAN something) has to reach "average Joe", not just the people on Hard Forum)
 
Last edited:
Dario D. said:
The average computer user isn't excited about SSD's, sees NO real reason to buy one (in fact, the price + lame storage space repels people), and so if anything, the HDD makers are killing the public perception of the "brand" of SSD's - something that could be devastating, if they don't do something quick to give them some zazz. (in fact, if I were an HDD maker, I wouldn't even call them "SSD's" - what kind of retarded, cryptic phrase is "Solid State Drive"?... I'd call them Ray Drives or something, and bundle them in video-card style boxes, and of COURSE make them powerful enough to keep their buzz alive longer than a day. I mean, word about the these things (speed/space/price that all MEAN something) has to reach "average Joe", not just the people on Hard Forum)
Click to expand...

The average joe isn't excited about hard drives either. They just buy them because they're Good Enough. Today, one look at the price of SSD is enough to turn away all buyers looking just for storage.

Give it time and SSDs will be adopted. Not because people actually want them; the average user could care less what's in their box and no amount of marketing will fix this. Rather, it's because the price has trickled down to the point where the intended application can benefit from SSD without the major pitfall of price. Today, some netbooks already reached that point with small capacity SSDs taking the place of a hard drive. In a few years time, perhaps even Dell and the like will have SSDs standard in mainstream computers.

When looked at from this perspective, people just don't want a dual head/actuator hard drive. Standard hard drives are good enough for the average joe, people with need for mass storage with high sequential performance have RAID, and the enthusiasts have SSDs - which will eventually (soon?) trickle down and become more accessible to the masses.
 
Last edited:
I understand what you're saying, but the way super-fast hard drives would be marketed to "average Joe" (whether SSD's or multi-head hard drives) would be with a giant sticker on the box that says:
"Programs open in half the time!
Windows boots 2x faster!
Files transfer at lightning speed!"

It's true that CURRENT hard drives have no mainstream appeal other than storage, but, by nature, I think companies would market super-fast hard drives specifically for their speed (not just storage)... and in a way that Average Joe can see (on the box). If they had commercials for this stuff (multiple companies chipping in to advertise SSD's or multi-head drives in general, for instance), it would be a heck of a lot easier, because then Average Joes would specically start seeking out computers/laptops that had these parts in them.
 
Aside from some marketing blurb on the retail box (which almost all products have anyway), an attempt at mass marketing would only make sense for things that have a relatively decent profit margin. Companies wouldn't want to try marketing items with razor-thin profit - especially something like a hard drive since those aren't "sexy". This is also why you generally won't see TV commercials or magazine ads for LCD monitors.

A company might increase sales but will have to increase price of the drive to compensate. However, when people are only looking at price the company stands to lose a good bit of sales against the cheaper competitors. Something like a dual actuator-head drive can't afford to be too expensive if it's to be attractive; and we just don't know how expensive these drives would be without the marketing. We don't even know how much of a performance gain there will be or if it'll be worth it.

Better to market something profitable like an iPod. They're popular, 'sexy', and profitable.

Now, I don't know how profitable SSDs are right now; maybe those can be effectively marketed toward gamers and enthusiasts while raking in a decent profit. However, that's not going to last forever; eventually drives will need to become cheap enough to be put mainstream - the same mainstream that buys Dell and doesn't particularly care about file transfer speeds. Marketing dollars will become wasted then.
 
Last edited:
SSDs are nice, but they still (mid-2014) lag considerably behind hard drives on capacity/cost, which is very important for many customers. A 256 GB SSD costs about the same as a 3 TB HDD. A factor of 12... very significant.

I disagree that additional arms/heads per platter would decrease reliability. I think it depends on the nature of the failure. More heads presumably means greater chance of a head crashing into the platter. But on the other hand, a "non-crash" failure of an arm/head would not prevent data from still being readable and writable, as long as at least one of the other arms/heads still functions. (Not at the higher speed available when more arms/heads are functioning, of course.)

It should be more reliable than a RAID-0 pair of drives. Together the pair have two spindles, which will fail more often than the single spindle of a drive with multiple arms/heads. (And a pair of "ordinary" drives have the same total number of arms/heads as a drive with two arms/heads so that term of the failure rate would be similar.) Also, a failure of one drive of a RAID-0 pair loses all data of both drives (hopefully backed up), whereas (as noted above) data on a drive with multiple arms/heads is still accessible as long while at least one of the arms/heads still functions.

Give a drive 3 or 4 arms/heads and sequential transfers will be much faster than with a RAID-0 pair of ordinary drives.

It should also be slightly faster at random access than a RAID array due to reduced average latency. (Latency is the time it takes for a requested sector to spin around until it passes underneath a read/write head.) With more read/write heads, a sector doesn't need to spin as far to reach the nearest head. Drives could be made to spin more slowly to save energy and reduce heat & noise, with only a small reduction in performance.

There might be room for 4 arms/heads without increasing the form factor, by placing the pivoting base of each arm, with its servo motor, within a corner of the "squared" circle. (The length of each side of the square equals the diameter of the platter, so the length and width of the overall drive wouldn't increase, assuming the base of the arm fits within a corner of the square.)

I assume hard drive manufacturers believe they will sell more drives (for high performance RAID arrays) if they don't offer high performance drives with extra arms/heads. Aren't there alternatives to unfettered capitalism? :) Maybe a tax on drive spindles and a credit for read/write heads. Maybe a law that requires vendors to accept packaging and failed products from their customers, to keep them from being disposed of in landfills and to give vendors an incentive to minimize the total that needs to be disposed.
 
The market for such a drive would be too small making the cost high.
 
Last edited:
What I'd really like to see is a 3.5" SSD. Imagine how much storage could be crammed into the 3.5" Form. With 1 TB SSD now, I could see 12TB in 3.5" form easily. And would still fit in all cases now.
 
SockMan! said:
In the case of magnetic media, perhaps research into Hard Rectangular Drives will be more worthwhile?
Click to expand...

I was fascinated by that article until I noticed the date. 2009. Five years ago, or multiple generations of disk drive product. IF these guys were destined to succeed, they probably would have done so already.

Venture capitalists aren't infinitely patient or foolish with their money, particularly after 2008. In fact, if you go to this page, you will see that this company is has received only "seed money" grants. I have to think that these guys pitched any number of VC firms, who declined to do an investment even in a Series A round where the company would still have a very low valuation.

x509
 
Solarnet said:
What I'd really like to see is a 3.5" SSD. Imagine how much storage could be crammed into the 3.5" Form. With 1 TB SSD now, I could see 12TB in 3.5" form easily. And would still fit in all cases now.
Click to expand...

PCIe is where it is heading due to SATA limitations.
 
SSD's are where its at. Storage/price has come down greatly and I have "made due" with 256Gb drives in my laptop and desktop for quite a few years now and before that I found that 80Gb was OK. If new tech should be developed it needs to be FASTER than SSD are store far more than spinners.
 
mkrohn said:
SSD's are where its at. Storage/price has come down greatly and I have "made due" with 256Gb drives in my laptop and desktop for quite a few years now and before that I found that 80Gb was OK. If new tech should be developed it needs to be FASTER than SSD are store far more than spinners.
Click to expand...
I wish they would make 3.5"-sized SSDs.
 
Cerulean said:
I wish they would make 3.5"-sized SSDs.
Click to expand...

I can see this for the enterprise but consumers would not want to pay $5000 for a 10 TB SSD or even $2500 for a 5TB model. My point is the form factor is not really the limiting factor on drive size.
 
Last edited:
drescherjm said:
The market for such a drive would be too small making the cost high.
Click to expand...

Can you provide a link to market research that supports that claim? Or at least state your assumptions? Suppose it would add $50 per drive to increase the arms/heads from 1 to 4 if the market volume is large. Why wouldn't the market be large for large hard drives (say 3 or more TB) with 600 MBytes/sec transfer rate that cost only $50 more than same-size drives with 150 MBytes/sec transfer rate? How small would the price differential need to be for there to be a large market for such ultra-fast drives?

Shall we agree to call a multi-headed hard drive a Hydra? Hail Hydra!

Question for all: Would most of the theoretical maximum performance of a Hydra be achieved if all heads are forced to seek to the same cylinder? I think that symmetry of motion would minimize vibration (assuming the arms/heads are spaced equally around the platter). And it would simplify the read/write algorithms. (Higher performance could be achieved later with smarter firmware.)
 
Suppose it would add $50 per drive to increase the arms/heads from 1 to 4 if the market volume is large.
Click to expand...

I do not believe the cost of the parts will be the biggest expense. I think it would be the lack of demand. This would be a much smaller market than a regular hard drive.

with 600 MBytes/sec transfer rate that cost only $50 more than same-size drives with 150 MBytes/sec transfer rate?
Click to expand...

What about 4 times the cost of a consumer drive instead of just $50 more?

Why wouldn't the market be large for large hard drives (say 3 or more TB) with 600 MBytes/sec transfer rate that cost only $50 more than same-size drives with 150 MBytes/sec transfer rate?
Click to expand...

Would an OEM put these in a PC? Would these be in USB3 externals?
 
Last edited:
keenan said:
I was thinking the OP was referring to staggered heads on the same armature, each reading offset tracks. That way you could read 2 or 3 parallel tracks simultaneously, and with the large buffers on today's drives, the drive firmware could then reassemble these bits into their proper order. I can see how this could increase sequential performance a good deal, but it obviously won't help random performance at all. Still, it seems it could be worthwhile - there must be a technical reason why this doesn't work.
Click to expand...

Things move around too much, unlikely all the heads would line up with the correct tracks at the same time.
 
bexamous said:
Things move around too much, unlikely all the heads would line up with the correct tracks at the same time.
Click to expand...

Agreed. This would not work at all because it would be impossible to correct for the variation between the tracks all at the same time.
 
If you put another head/motor system on the opposite side, it could allow for improved concurrent access. Like if you were doing a sequential transfer with one head, the second head could be accessing other smaller files. It would slightly ease the Achilles heal of platter drives.

Basically something like this.

double_drive.jpg
 
You're making a fancier NCQ for closer to double the cost and failure rate. How about just double the last few bits and have two HDDs. HDD is just not best place to put complexity, let a fancy FS/RAID controller figure it out.
 
I buy tons of hard drives. I don't even look at the performance numbers anymore. They're plenty fast enough, even the slowest ones (in 3TB and 4TB sizes), for storage. Anything that requires speed is on SSDs.

MrGuvernment said:
PCIe is where it is heading due to SATA limitations.
Click to expand...

PCIe is an interface, 3,5" is a form factor. PCIe has plenty of connectors and cables and drives will definitely stay 2,5"/3,5". Look up SATA Express.
 
You must log in or register to reply here.
Back
Top