Revenge of the magnetic storage medium

[WildW]

Saw this story about a new kind of magnetic hard disk technology that will supposedly make NAND Flash look slow. . . . sounds promising. I love living in the future


DataSlide reinvents hard drive - 64 parallel read-write heads

 

[SunSamurai]

From the title alone I can tell thats a huge FAIL. 64 r/w heads is asking for a drive to screw up.

 

[dighn]

Originally posted by: aeternitas
From the title alone I can tell thats a huge FAIL. 64 r/w heads is asking for a drive to screw up.

the heads are non-moving. that by itself should be a big increase in reliability.

pretty interesting technology. very similar to existing HDDs, but more "solid state".

 

[taltamir]

an ultra-thin, 2-dimensional array of 64 read-write heads, operating in parallel, is positioned above an piezo-electric-driven oscillating rectangular recording surface

Ok, I am a bit confused on how it works.
I am getting that it is a rectangle with 64 write heads.... it does not spin... but... what? It sounds like the rectangle is moving back and forth which is fail, it makes more sense to move the heads.
More likely though it is just a row of 64 heads placed on a rectangle platter, and they move back and forth across it. This sounds less durable than a regular drive with multiple heads per arm, but time will tell.

anyways, lab speed of a sample and a mass producible product are two different things. How fast will SSD drives be when this thing hits the market?

 

[dighn]

Originally posted by: taltamir

an ultra-thin, 2-dimensional array of 64 read-write heads, operating in parallel, is positioned above an piezo-electric-driven oscillating rectangular recording surface

Ok, I am a bit confused on how it works.
I am getting that it is a rectangle with 64 write heads.... it does not spin... but... what? It sounds like the rectangle is moving back and forth which is fail, it makes more sense to move the heads.
More likely though it is just a row of 64 heads placed on a rectangle platter, and they move back and forth across it. This sounds less durable than a regular drive with multiple heads per arm, but time will tell.

anyways, lab speed of a sample and a mass producible product are two different things. How fast will SSD drives be when this thing hits the market?

It does look like the heads are fixed, and it is the recording media that is oscillating. If that's the case I'm sure they have a good reason for doing that. It might be that designing so many heads/connections to be oscillated at high frequencies is difficult.

How feasible this technology is is a question. But I think it definitely has some advantages, e.g. it's very similar to existing HDDs so a lot of the technologies involved can probably be reused, and there are no massive arrays of tiny components so manufacturing might be less demanding etc.

 

[aka1nas]

Who cares about the transfer rates? While the access time is probably less than spinning platter drives, it still has mechanical parts and will have much slower access time than an SSD.

Even if the thing could do a few GB/sec in sequential transfers, what interface are you going to hook it up to take advantage of it? PCI-E x16 is a non-starter, as few server boards have multiple x16s. In the consumer space, you're pretty much limited to the top few % who buy SLI/Crossfire boards and somewhat unsurprisingly tend to populate their x16 slots with GPUs. x8 and x4 slots are practically non-existent, and with an x1 slot you might as well just use SATA.

SATA III is only another doubling of transfer rates, so that only takes us to 600 MB/sec still.

Edit: Actually read the article now. In what world does a device with a 500MB/sec transfer rate saturate a "32-lane" PCI-E bus?

 

[taltamir]

it boasts really high IOPS combined with no limit on writes. the target market seems to be servers that need high transaction rate.