Writers wanted - Previous experience not necessary, you only need an interest and we can help every step of the way. Please PM qubit or alexp999 for details.

Jump to content

Welcome to techngaming.com
Register now to gain access to all of our features. Once registered and logged in, you will be able to create topics, post replies to existing threads, give reputation to your fellow members, get your own private messenger, post status updates, manage your profile and so much more. This message will be removed once you have signed in.
Login to Account Create an Account


Advertisement


Newest Downloads


Today's Wallpaper


- - - - -

[PR] Seagate's 1 Terabit Per Square Inch Milestone for HDDs Demo


Click image to enlarge



Advertisement
CUPERTINO, Calif. - March 19, 2012 - Seagate (NASDAQ:STX) has become the first hard drive maker to achieve the milestone storage density of 1 terabit (1 trillion bits) per square inch, producing a demonstration of the technology that promises to double the storage capacity of today’s hard drives upon its introduction later this decade and give rise to 3.5-inch hard drives with an extraordinary capacity of up to 60 terabytes over the 10 years that follow. The bits within a square inch of disk space, at the new milestone, far outnumber stars in the Milky Way, which astronomers put between 200 billion and 400 billion.

Seagate reached the landmark data density with heat-assisted magnetic recording (HAMR), the next- generation recording technology. The current hard drive technology, Perpendicular Magnetic Recording (PMR), is used to record the spectrum of digitized data – from music, photos, and video stored on home desktop and laptop PCs to business information housed in sprawling data centers – on the spinning platters inside every hard drive. PMR technology was introduced in 2006 to replace longitudinal recording, a method in place since the advent of hard drives for computer storage in 1956, and is expected to reach its capacity limit near 1 terabit per square inch in the next few years.

“The growth of social media, search engines, cloud computing, rich media and other data-hungry applications continues to stoke demand for ever greater storage capacity,” said Mark Re, senior vice president of Heads and Media Research and Development at Seagate. “Hard disk drive innovations like HAMR will be a key enabler of the development of even more data-intense applications in the future, extending the ways businesses and consumers worldwide use, manage and store digital content.”

Hard drive manufacturers increase areal density and capacity by shrinking a platter’s data bits to pack more within each square inch of disk space. They also tighten the data tracks, the concentric circles on the disk’s surface that anchor the bits. The key to areal density gains is to do both without disruptions to the bits’ magnetization, a phenomenon that can garble data. Using HAMR technology, Seagate has achieved a linear bit density of about 2 million bits per inch, once thought impossible, resulting in a data density of just over 1 trillion bits, or 1 terabit, per square inch – 55 percent higher than today’s areal density ceiling of 620 gigabits per square inch.

The maximum capacity of today’s 3.5-inch hard drives is 3 terabytes (TB), at about 620 gigabits per square inch, while 2.5-inch drives top out at 750 gigabytes (GB), or roughly 500 gigabits per square inch. The first generation of HAMR drives, at just over 1 terabit per square inch, will likely more than double these capacities – to 6TB for 3.5-inch drives and 2TB for 2.5-inch models. The technology offers a scale of capacity growth never before possible, with a theoretical areal density limit ranging from 5 to 10 terabits per square inch – 30TB to 60TB for 3.5-inch drives and 10TB to 20TB for 2.5-inch drives.

The 1 terabit per square inch demonstration extends a long line of storied technology firsts for Seagate, including:
  • 1980: ST-506, the first hard drive, at 5.25 inches, small enough to be widely deployed in early microcomputers, the precursor of the modern PC. The 5 megabyte drive cost $1,500.
  • 1992: The first 7200RPM hard drive, a Barracuda ® drive
  • 1996: The first 10,000RPM hard drive, a Cheetah ® drive
  • 2000: The first 15,000RPM drive, also a Cheetah hard drive
  • 2006: Momentus ® 5400.3 drive, a 2.5-inch laptop drive and the world’s first drive to feature perpendicular magnetic recording technology
  • 2007: Momentus FDE (Full Disk Encryption) drive, the industry’s first self-encrypting hard drive
  • 2010: Momentus XT drive, the first solid state hybrid hard drive, combining traditional spinning media with NAND flash, to deliver speeds rivaling solid state drives (SSDs)
Seagate achieved the 1 terabit per square inch breakthroughs in materials science and near-field optics at its heads and media research and development centers in Bloomington, Minnesota, and Fremont, California.
http://seagate.com/twitter
http://seagate.com/facebook
http://media.seagate.com


9 Comments

This is why HDDs will continue to compete against SSDs. You just can't get the price to capacity ratio with nand storage. I just hope HDD prices continue to come down to at least close to what they were.
  • Quote
Photo
Static~Charge
Mar 20 2012 06:13 PM
Hard drive capacities are good as they are now. I wish the manufacturers would devote some time and effort into improving their reliability.
  • Quote
Improve reliability and transfer rate, I say. It takes hours to image one 2GB drive to another, for example.

EDIT: I meant 2TB. Sorry. :blush:
  • Quote

Improve reliability and transfer rate, I say. It takes hours to image one 2GB drive to another, for example.


Do you mean 2TB?
  • Quote
Oh duh! :facepalm: Ya, 2TB. I knew something didn't look right lol.
  • Quote
Transfer rates don't get much better on this technology other than with integrated digital memory - "hybrid" drives, which are basically cheating by using seriously huge caches, so don't improve speed for all workloads.

There is room to improve reliability - that's what they've done with enterprise grade drives, typically SAS, the kind of drives used in data centres. The only reason we end users don't buy them is that they're both smaller capacity and, most importantly, just so prohibitavely expensive. The reason they're so expensive is because of all the extra work and parts involved in giving them that extra reliability. Almost all the components are higher quality with > MTBF. There are sensors to detect minute vibrations and make instantaneous adjustments to the read head's position. There's ECC everywhere so errors, wherever they may be introduced, are detected - and then they're even corrected faster than desktop drives. Yet, even enterprise drives fail at a fair rate - the very nature of what we're doing makes regular failures inevitable, we currently just try to handle them acceptably so end users don't (often) notice. Hence internal ECC, spare sector pools, RAID5, RAID1, ZFS RAIDZ, ZFS's ECC codes, scrub, fsck, sfv, parity files, local/external backups, SRDF, DRBD, remote backups, cloud backups, ... SpinRite for when it all goes wrong... ;) Unforunately, as you can see, errors in long-term storage are and probably will be a fact of life for as long as we continue using mechanical devices. And if you want greater reliability, you have to pay a lot more money. So the typical consumer approach is to double the drives with RAID1 rather than use more reliable drives - it's cheaper.

(Note: RAM suffers from problems too: occasional EM interference does flip bits. Very very infrequently, granted, but it happens. Know how much ECC we end users typically have going on there? None. We assume RAM is 100% perfect 100% of the time. Enterprise does use ECC RAM though.)

The way forward is to change the technology from PMR. Heat-assisted is probably the next one to arrive as q has described, and holographics I'm guessing will come a few years after; and all these are aimed at increasing capacity. That's the biggest issue at present. We've been stuck at 3TB (consumer) for some time; what with blu-ray movies (50GB per disc), HD youtube (GBs), FLAC (30MB per song), 1080p+ camcorders (>> GB/TBs), 20MP+ cameras (60MB per photo) - 3TB isn't all that much considering these sort of loads, especially when many people have 50% storage as redundant storage. As an example, my RAIDZ pool is currently on 33% redundancy, then I have some RAID1 configs obviously on 50%. HA data will often have 75% redundancy or more by virtue of a RAID1-esque setup on one box, then a remote duplication link to another site also with a RAID1-esque setup. They might then take periodic offline backups of that. And that's on small expensive enterprise grade drives!
  • Quote
Just to make things more tangible & interesting (yawn...), some scribbles.

Let's say you have just 4TB of data that you want to store (not that much really if you think about it). As an end user, you could buy a 3TB (~£160) and a 1TB (~£90), so the total would be £250 and you'd have your 4TB. Now let's consider you're a business storing very important data and you needed enterprise grade HA storage.

500GB 2.5" enterprise grade SAS, ~ £250.

How many drives would you need? 4000/500 = 8 drives. That's £2,000, so 8x the end user cost.

But you never fill a pool up to 100% - 80% is the max recommended for most FSs. Plus you need room to expand. So you need say 8TB of storage. that's 16 drives. But you don't get 1TB available storage on a 1TB drive, do you? (1000 vs 1024). A 500GB would give you about 465GB. Then minus some more for the FS's PTs, inode tables and suchlike... let's say you're down to 450GB. 8000/450 = 17.7. So we would need 18 drives to get 8TB actual. That's £4,500.

Oh yes, HA... so we need a local RAID1. We have to double up to 36 drives. That's £9,000.

And now for replication to a remote mirror. We have to double up again. That's 72 drives. £18,000.

How about a periodic offline backup? x1? x1.5 to allow for full + some diffs?

How much do you think the I/O cards cost to support 36 drives in a box? Or the rest of the server hardware? Or the remote replication? Or the support contracts with the various suppliers? What about costs of replacement drives for hotswapping failures? The wages for the infra teams to maintain all these 72 drives and these servers?

All this for 4TB which you could buy for £250... So you see, capacity does matter :)
  • Quote
Photo
Static~Charge
Mar 21 2012 01:32 PM

Oh yes, HA... so we need a local RAID1. We have to double up to 36 drives. That's £9,000.


RAID 1 for 8TB of usable disk space? That's insane (and expensive). RAID 6 would be a better choice. Instead of doubling the number of drives, you just add 2 more. If you're extra paranoid, add another drive as a hot spare.
    • IlluminAce likes this
  • Quote
The implementation you choose depends on precise requirements and workloads, and the priorities and decisions on those in infra, management and finance. The point I was trying to get across is that storage capacity is still a major factor where improvements would be very much welcomed by consumers and, more importantly, businesses.

With regard to the example, there are a whole raft of possibile implementations - standard RAID configs, filesystems, custom RAID and plenty of rather expensive third party solutions too. I agree that RAID6 is a nice option and offers something almost comparable to RAID-Z2. However, if you're running a write-heavy workload then it's probably not your best bet. And array rebuild will have nicely distributed I/O, but be very computationally intensive on the RAID controller. All drives attached to it will be affected, and the rebuild will take quite some time. Finally it's true that it only requires +2 drives, but you wouldn't want to run all that many drives together in a RAID6 set, due to those rebuild times and the performance impact thereof. Maybe the business might not be so concerned about the rebuild as they could delay it until a period of light utilisation (RAID5 in the meantime), or they'd switch to the secondary whilst the rebuild is in process. So it's a very different solution to RAID1/1+0 with different considerations. Personally I'm a ZFS convert, but that has its own ups and downs :)
  • Quote
or Sign In

Please Note - To prevent spam, all comments must be approved by a moderator. Registered members should discuss this article in the Forums