Seagate's Shingled Magnetic Recording to bring 5 TB HDDs in 2014

Scorpus

Posts: 2,162   +239
Staff member

seagate shingled magnetic recording hdds hdd storage hard drive hard disk smr

Seagate has tried its hardest to shrink the read/write heads of hard drives even further, but with the latest 1 TB 3.5-inch platters, the company believes they've hit a physical brick wall: the head simply can't get any smaller. The same goes for the tracks on the platters, which Seagate claims are as close together as physically possible.

Essentially this means that without increasing the amount of platters in a drive, producing drives with capacities above 4 TB (four 1 TB platters) has become rather hard. To get around these physical limits, Seagate has looked into changing the architecture implemented, coming up with Shingled Magnetic Recording (SMR).

Hard drive tracks are typically the size of the write heads, as they are larger than the read heads. However the only part of the track that the read head actually cares about is the central data band; there is guard space the width of the write head either side, which is there to reduce the chance of reading adjacent tracks. SMR takes advantage of this, overlapping tracks to reduce the guard space, allowing areal density to scale without the need for shrinking read/write heads.

A downside to implementing SMR on a hard drive concerns writing data to a series of tracks that already contain some data. As the tracks overlap, and the write head is now the width of more than one track, writing to one track can destroy the data in an adjacent track. This results in all subsequent tracks needing to be re-written, which reduces performance.

Seagate's SMR design groups tracks into bands, with each band breaking the "shingled" layout so that the entire disk doesn't need to be re-written after writing to just one track. That said, some re-writing will still need to occur, and it's not clear exactly what effect that has on drive speed compared to non-SMR disks.

Thanks to the technology, Seagate are hoping to increase 3.5-inch platter density from 1 TB to 1.25 TB, paving the way for 4-platter, 5 TB SMR drives in 2014. Apparently Seagate has already shipped one million SMR-enabled drives, and are hoping that by increasing the platter density, performance is not degraded compared to their existing SMR HDDs.

Permalink to story.

 
Meanwhile, certain companies are quietly leaving Seagate and others in the dust.
newegg.com/Product/Product.aspx?Item=N82E16820147251
 
Meanwhile, certain companies are quietly leaving Seagate and others in the dust.
newegg.com/Product/Product.aspx?Item=N82E16820147251

There's still a large market for traditional harddrives. At $850.00 The vast majority of consumers are not going to by that 1TB harddrive. I have a feeling seagate is also working hard on SSD's
 
So let me get this straight - they have to do a ton of rewriting of data when a new track is laid because you are partially destroying each of the adjacent tracks so they get re-written which destroys the data on the other side of that track and so on and so on. The only reason that they ENTIRE disk doesn't need to be rewritten is they put in bands that break the chain.

...each band breaking the "shingled" layout so that the entire disk doesn't need to be re-written after writing to just one track.

That sounds like a huge hit to performance just to achieve a modest 25% increase in capacity. Doesn't sound worth it.
 
Ranger12 said:
Guest said:
Meanwhile, certain companies are quietly leaving Seagate and others in the dust.

newegg.com/Product/Product.aspx?Item=N82E16820147251

There's still a large market for traditional harddrives. At $850.00 The vast majority of consumers are not going to by that 1TB harddrive. I have a feeling seagate is also working hard on SSD's

Agreed. In fact, you can get a Samsung 128GB SSD and four Seagate 4TB HDD for the same price which gives you a much better value.
 
I read an article a little while ago from a Seagate executive that explained that solid state drives could not replace traditional disk hard drives any time soon. His logic was based on the manufacturing capacity, investment requirements, and technical density limits for solid state manufacturers to increase their output. Combined with the growing storage requirements, we will be using traditional disk hard drives for quite some time. I seem to remember some line in the article like if every chip fab in the world started to only produce solid state drive chips they would only be able to hit some low single digit percentage of the annual storage requirements. I sure can't afford to store all my DSLR images on solid state drives, and I just keep taking more and more. Just wait until every cell phone is shooting ultra 4k high def video. Do you think Google and Amazon use traditional disk or solid state hard drives for the vast majority of their storage needs?
 
Why are they bothering with such density increases when all those 5.25 bays in our computers are just sitting there wasted.

Without having to invest any extra money in new tech they can just release a giant *** 5.25 drive that instantly gives you 50% or more storage with no hit to performance and since the bay is higher you could even stack more platters.

Bring back the quantum bigfoot! http://en.wikipedia.org/wiki/Quantum_Bigfoot_(hard_drive
 
This will only really come in handy in the server sector. 4TB is more than enough for most people. Even professionals like people who use Photoshop and need scratch drives usually want multiple hard drives as opposed to one with a large capacity.
 
From the point of view of information theory, there is a way to increase e.g. HDD capacity without decreasing size of magnetic dot: allow dots to overlap - by more precise positioning of dots.
Assuming that overlapping dots cannot be simultaneously magnetized, this constraint reduces entropy per site, but there can be more sites.
For example rescaling 2D lattice by sqrt(2) we get twice more sites, but there cannot be used "1" in two neighboring sites (so called hard-square model) - the entropy drops to about 0.588 bits/node because of this constraint. As there is twice more nodes, we get about 2*0.588=1.176 bits per old node: 17% capacity increase (arxiv:0710.3861).
Theoretically we could increase capacity a few times this way. However, handling such medium would be too complicated for real-time application, but maybe for backups etc. ...
 
Meanwhile, certain companies are quietly leaving Seagate and others in the dust.
newegg.com/Product/Product.aspx?Item=N82E16820147251

Gee, what if there were people who need massive amounts of storage at an affordable price. Like people who work with large amounts of data, or data centers. Last I checked, a 7200rpm 1TB drive is like 9x as cheape than the cheapest 1TB SSD. Surely there's a market for them.
 
Guest said:
Ranger...that 1TB SSD is $599 not $850.

Actually, it's $649, not $599. If it was $599 then Newegg must have changed its price twice and I say it twice because it was $850 this morning.
 
"Meanwhile, certain companies are quietly leaving Seagate and others in the dust.
newegg.com/Product/Product.aspx?Item=N82E16820147251"

WHATTT!?! even at $649 I think I'm game!
Did I miss a post about this on TechSpot?
 
Why are they bothering with such density increases when all those 5.25 bays in our computers are just sitting there wasted.

Without having to invest any extra money in new tech they can just release a giant *** 5.25 drive that instantly gives you 50% or more storage with no hit to performance and since the bay is higher you could even stack more platters.
I know, right? I have 5x 5.25" bays, but am only using one of them for a DVD-RW drive. I'd love some high-capacity drives, even if they'd consume a lot more power and cost more.
 
Back