Sony researchers have succeeded to increase the capacity of magnetic tape by increasing the areal data density by 74 times. This could make 185 TB tape cartridges a reality and can be compared to the latest generation LTO-6 (Linear Tape-Open) that has a density of 310 gigabits per square centimeter, or 2.5 TB uncompressed data per cartridge. There's more:
Used for storage since the first digital computers, magnetic tape has been eclipsed by hard disk drives and flash drives as a medium in recent years but is still in use to preserve critical information over the long term in data centers, corporate archives and other facilities.
To make the new recording material, Sony used a kind of vacuum thin film-forming technology called sputter deposition. The process involves shooting argon ions at a polymer film substrate, which produces layers of magnetic crystal particles. By tweaking the sputter conditions and developing a soft magnetic underlayer on the film, the manufacturer was able to create a layer of fine magnetic particles with an average size of 7.7 nanometers.
At the Intermag Europe 2014 international magnetics conference starting in Germany on May 4, Sony will describe the new technology in a presentation with IBM, which helped measure the new density. Sony said it wants to advance the thin-layer deposition technologies and commercialize the new tape, but it did not say when such a product could appear on the market.
(Score: 4, Funny) by bob_super on Friday May 02 2014, @05:39PM
How many libraries of congress, or how many TV series marathons, was that again?
(Score: 2) by fishybell on Friday May 02 2014, @05:43PM
Hell, even if they artificially slowed the release rate to 10 tB on first release, 20 tB on first upgrade, 40 tB on second upgrade, etc. I'd still see this as a huge advantage over LTO-6. Being able to back a medium sized SAN's worth of data on a few tapes would be vastly superior to backing up to another SAN for archiving. If you don't need the data back more than once in a blue moon, tapes are definitely the way to go.
(Score: 2) by Hairyfeet on Friday May 02 2014, @07:07PM
This is great for the enterprise, but what about SMBs and home users? We really need a new affordable long term storage solution, DVDs are too small and I haven't seen any real tests on BR used as long term storage and with HDD sizes settling in the 1-4 TB range (price jumps insanely after you go past 2 TB) and not being really designed to sit idle for long periods we really need an affordable long term solution for the SMB and home markets. Folks have more they need backed up every day, hell all the HD cameras and videocams are causing many of my customers to upgrade their drives but other than portadrives there just isn't anything I can point to and say "This is a good long term solution".
So while I'm happy the enterprise guys are getting new storage i really wish some of this stuff would trickle down to us in the trenches. A pair of say 4-6 TB tapes would probably be the perfect backup solution, run one for a couple weeks and then swap it for the other which is stored in a safety deposit box but the drives are just crazy high, and of course dragging around spinning rust increases risk a bump will kill the drives...sigh we really need something better.
ACs are never seen so don't bother. Always ready to show SJWs for the racists they are.
(Score: 2) by takyon on Friday May 02 2014, @09:08PM
The price jump isn't too insane for >2 TB. The price now apparently scales linearly to 5 TB at $160 [slickdeals.net]. 3 TB often goes down to $90-100. So we're looking at $30-32 per TB for the best price (and have been for years now). That just leaves the current 6 TB drives available to consumers, sold under the premium LaCie brand, and maybe Seagate too. And now that 5-6 TB are finally released and available to consumers, there's speculation of 8-10 TB coming soon [theregister.co.uk]. Those would probably require HAMR or shingles, though I could see WD/HGST pushing 8 TB on 7 platters without HAMR.
[SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]
(Score: 2) by Hairyfeet on Friday May 02 2014, @11:10PM
Uhhh you missed my point friend because HDDs are NOT MADE TO BE PARKED for long terms so are NOT a viable long term storage solution. While we have all heard about the "300MB drive that is still good" I can tell you there are 50 more times when you plug in that old drive and suffer a head crash.
No what i am talking about is long term and right now we just don't have a viable product. DVDs last a decent length but are too small, and BR drives and discs just haven't been tested AFAIK for long term storage. what we really need is something like tape for the home and SMB markets, something you can put everything from the family photo to tax records on.
ACs are never seen so don't bother. Always ready to show SJWs for the racists they are.
(Score: 3, Funny) by takyon on Saturday May 03 2014, @08:00AM
Let them eat cloud.
[SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]
(Score: 2) by Hairyfeet on Saturday May 03 2014, @03:24PM
You make a funny but I actually looked into that and not only has every major cloud service suffered one or more failures but the amount of turnover in that sector makes cloud useless for anything but VERY short term storage. Compare this to say DVD where I have 1x DVDs I burnt over a decade ago that are still 100% readable and you see why cloud just isn't a solution, in fact there really isn't ANY long term storage for the SMB and home users other than burning discs and sadly disc sizes just haven't kept up with data growth.
I have looked into BR but I'm afraid BR is crippled so badly by the DRM that the ability to playback the discs is brought into serious doubt, its obvious that unlike DVD Sony only cared about making BR MPAA friendly and secondary uses be damned. that is probably why computer usage of the format is so anemic and appears to be flatline, you just can't trust it, what with the constant whack a mole firmware upgrades to keep patching the anti-piracy crap.
ACs are never seen so don't bother. Always ready to show SJWs for the racists they are.
(Score: 2) by takyon on Saturday May 03 2014, @10:51PM
Maybe what you want is Sony/BR/Panasonic "Archival Disc" [wikipedia.org] (planned 300, 500, 1000 GB). But even that's "not intended for consumers".
[SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]
(Score: 2) by Hairyfeet on Sunday May 04 2014, @01:29AM
When you see "not intended for consumers" you should translate that as "starts at over a thousand, more likely several thousand and media at $20+ a pop minimum", no different than tape is now its just not affordable or consumer friendly in the least.
ACs are never seen so don't bother. Always ready to show SJWs for the racists they are.
(Score: 2) by edIII on Friday May 02 2014, @09:26PM
You outlined the real problem already. Trust in the medium.
I don't trust tapes. The potential for errors is too high and I don't see myself being able to trust that a file will be intact 5 years later. I just don't. Too much experience with floppy disks, old tape systems, Iomega's Jazdrive bullshit. While I'm sure it's impressive technology, I can't take a system for storing data on magnetic mediums seriously that aren't sealed systems. Tape just seems old, analog, and crappy IMHO. Give me actual memory all day long, or better yet, all that 3D optical crap we were promised 20 years ago.
What I do trust though is a ZFS cluster. That's the Cadillac of reliability since it constantly checks itself and can self repair before problems get too bad. It uses hard drives though. The question is what happy balance is there between reliability and cheapness? What product provides that?
If they want me to use that tape system I need to use more than one tape, or split the tape into two equal copies. At this point I am far more concerned about silent data corruption and bad DVDs than straight data loss. Straight data loss has happened to me less than 5 times in 30 years at the scale of an entire drive. Somebody deleting all the contents of share has happened a couple orders more, and silent corruption of a database or an excel spreadsheet many orders more than that. Thank god for daily backups and versioning systems.
I'm all on board with this tape system since it's so large, but that does mean it has to be considerably less expensive space versus a physical hard drive. While I could use it myself personally, most people will never see this in use.
What we need is a standalone tape based product that can create a dormant ZFS cluster that can self check and self repair, but do so maybe once a week and shut off. You can have data scheduled to be added to it with removable storage systems on offsite backups, and just a networking connection if you have local access to an active ZFS cluster.
These tapes alone don't represent a happy medium at all and are too large to mean anything to most sysadmins in small companies that might have 5 TB of actual business data and not porn backups from user profiles.
At least I wouldn't sleep well knowing that it's just on a tape sitting on a shelf in some office somewhere.
Technically, lunchtime is at any moment. It's just a wave function.
(Score: 2) by evilviper on Saturday May 03 2014, @01:37PM
"ZFS for tape" doesn't even make sense... You're confusing two completely different models of operation, and pretending that the problems of one are problems with the other, when they absolutely are not.
If you're afraid of silent data corruption, fine... You just have to pop your archive tapes into a drive once every X (eg. 2 months), and have it do a full read, while verifying the checksums of each file. CPIO has had built-in support for checksums on each individual file for I-don't-know how many decades, but a long damn time.
Obviously, if you throw your tapes in a warehouse, and leave them there for 20 years, you won't know when corruption happens, but ZFS won't help if it is used as an offline long-term storage file-system either.
And Z-RAID is nice for online disks, but tapes can do RAIT, whatever percentage of parity you want (par2), or just plain duplicate or heirarchical backups. I've used par2 to give data CDs redundancy, with 4 disks burned for every 3 CDs worth of data, and able to recover from an entire missing disk AND about 2% of block errors spread across the remaining 3 disks, or a massive number of errors (up to about 35%), in any distribution, across any or all the discs, if none of the set is a total loss...
So what do you think ZFS has to offer for tapes? Its modus operandi is giving hard drive arrays the features tapes have had (optionally) for 2+ decades before it came along.
Hydrogen cyanide is a delicious and necessary part of the human diet.
(Score: 2) by edIII on Saturday May 03 2014, @09:08PM
I'll give it a shot.
- Greater reliability - The whole point of performing backups in the first place is to achieve a specific level of reliability. ZFS provides much greater reliability than tapes alone. Putting multiple copies and using the methods you described are not really different than that of RAID, and ZFS actually solves some of the bigger reliability issues with RAID. None of what you mentioned solves the well known issues of RAID and is why we scream at people that RAID is not a backup. Using RAID methods against a tape does not solve the issues either.
- Why recreate the wheel? - ZFS already accomplishes much of what your data protection method aims to achieve and does it better. Instead of two projects needing experienced contributors to keep two different platforms doing the same work, you just have a smaller project designed to present the tapes to the ZFS code and provide basic administrative functions against a tape held snapshot. Don't settle for just ZFS. Any decent equivalent can be made to use tapes in the same way.
- Easier implementation - Just insert your tapes into the server and your existing ZFS snapshot that is available over the network or from an attached SAN with removable drives. This should be fairly simple to create. A 4U server to hold the tape readers and more than likely a SAN to populate with drives. Almost anybody here can build and operate that server, assuming the code was available.
- Easier administration - You're already familiar with ZFS and use it for your enterprise storage on a daily basis (an assumption I know). The tape system is not functionally different. It may need to be turned on to access, and has a greater lag time, but presents distinct advantages providing valuable tools. Such a system would not need to read the tapes and could use a database cached on local SSDs to review file versions and basic metadata. This allows an administrator to traverse directories and perform searches against the tapes with near instant results versus a significant wait time to search 185TB.
- Flexible schedule - You already wrote it using ZFS and can be confident that it is held safely. You can reinspect the values on the tapes once a week if you are very paranoid, or once every two months as you suggest.
- More confidence - Accessing the server's front panel can even visually show you the average ages of a checksum. Decide on a full check whenever you want. Much better than an assumption that the tapes are okay and the time it would take to check them to be sure. You already know from looking at the front of the server it has been 37 days since the last success and there have been 12 errors. Just walking past the server on a daily basis you know with a good deal of confidence exactly what's up. I don't know about you, but that just makes me feel better thinking about it. That last part of course is just a LCD programmed by the server every so often. You could make that with an Arduino and would be low power consumption.
- File history - Instead of just having redundant copies in case of failure, you have multiple versions to assist with data corruption. It's just there and works. No extra coding effort and keeping track of what datasets have what and when. With large capacities you can keep many versions.
- Ridiculous pool capacity - ZFS offers a way to partition, use, and administrate such a large space efficiently. Do you want to create some sort of directory and date based system to keep track of multiple sets manually? There may be backup administration software that tries to help with that. Probably is, but ZFS is also a pretty good contender to managing the whole thing with a growing number of 3rd party solutions and support, and some of it open source or FOSS. It's not some typically crappy backup software manufacturers attempt to maintain approximately whenever.
That was my best shot. Let me know.
Technically, lunchtime is at any moment. It's just a wave function.
(Score: 2) by evilviper on Monday May 05 2014, @03:34AM
Utter and complete nonsense. You can't name one thing ZFS on tape would give you. You clearly read nothing I said, and are incredibly ignorant of (very simple) tape functionality. But I don't have the time nor inclination to pound the point into your head.
Hydrogen cyanide is a delicious and necessary part of the human diet.
(Score: 2) by edIII on Monday May 05 2014, @04:13AM
I actually just did in that one sentence you quoted. You yourself claimed that the only thing providing data protection on the tapes was using RAIT, which is a form of RAID specifically for tapes instead of drives.
RAID is unarguably not a backup. You obviously consider tapes to be a form of backup. I do not. They are just single copies sitting on a questionable medium (my opinion) and forgotten. Just hope for the best and screw it. That's not a proper disaster recovery model to be following, especially in this day and age where databases and customer information is everything, and uptime sells contracts. Putting an additional copy on the same medium that will be stored on the same shelf is just ridiculously funny to me. You go through all the hassle and expense of a tape system and don't use a 2nd tape stored in a different facility.
You bring up RAIT, since it also unarguably provides RAID features and greater reliability. So if you obviously believe that RAIT is valuable and solves issues NOT PRESENT in tapes alone, then you yourself just stated the following:
By the very same logic, ZFS (the technology in general, not project specific), quite successfully solves many issues present with RAID, and therefore RAIT. The following is held to be true by logic alone:
Unless you just want to be hostile for some reason and make the claim that ZFS provides nothing of value over RAID?
It makes no difference to me that is spooled magnetic tape or sealed platters. You can apply RAID/RAIT all you want to them. That logically does not preclude ZFS from adding value to a RAID system, and RAIT is not that functionally different from RAID although that should be completely obvious to anyone.
Yeah, complete and utter nonsense. I have no idea why ZFS would be thought to have value in any kind of disaster recovery plan or storage system at all. I don't know what I was thinking....
Technically, lunchtime is at any moment. It's just a wave function.
(Score: 2) by evilviper on Monday May 05 2014, @03:27PM
CPIO provides checksums, just like ZFS.
ZFS DOES NOT PROVIDE ANY DATA PROTECTION.
Z-RAID provides redundant copies, so that a file that fails checksum can be copied from a different drive. Z-RAID provides the data protection, just like RAIT (or PAR2) can and does. RAIT and PAR2 is not necessary, and not commonly done, because just doing regular backups of the data and keeping multiple tapes provides all the redundant copies you could want
Read the above over and over again.
I still can't understand your response as anything but massive ignorance of the technologies you're talking about, and horribly wrong-headed conclusions based on that confusion.
Hydrogen cyanide is a delicious and necessary part of the human diet.
(Score: 2, Insightful) by redneckmother on Friday May 02 2014, @05:44PM
What part of LTO didn't Sony get? Another proprietary format, destined for the dustbin.
Mas cerveza por favor.
(Score: 2) by Nerdfest on Friday May 02 2014, @08:14PM
More likely destined for people who don't remember history, destined to be boned ... again.
(Score: 2) by Bartman12345 on Friday May 02 2014, @05:48PM
... what kind of read/write speeds can it sustain? What about seek times, 185TB is a lot of data to sift through sequentially to get to that one important file. Fitting that much data on a tape cartridge is way impressive, but capacity isn't everything.
(Score: 2) by Boxzy on Friday May 02 2014, @05:59PM
Access speeds, yes very important. Just as important is price, as it stands LTO is ten times the price of hard drives until you need around 50 tapes. I can buy new hard drives every three/four years over 10 years and still save half the price of a $2000 LTO drive and enough tapes.
Go green, Go Soylent.
(Score: 3, Interesting) by Grishnakh on Friday May 02 2014, @06:03PM
You hit on the most important factor: price. 185TB tapes might make sense for big datacenters doing archival backups, but for smaller businesses and consumers, it's overkill, and probably way too expensive if LTO* is anything to go by. LTO tapes are very reasonably-priced, but the drives are ridiculous, and also require a SAS interface, which only servers have.
What we need isn't 100+TB tapes, what we need is a 1TB/tape technology with a drive that fits in a 5.25" drive bay, connects with SATA, and costs less than $200.
(Score: 2) by Jaruzel on Friday May 02 2014, @08:49PM
What we need isn't 100+TB tapes, what we need is a 1TB/tape technology with a drive that fits in a 5.25" drive bay, connects with SATA, and costs less than $200.
What, like these ones [rdxworks.com]?
-Jar
This is my opinion, there are many others, but this one is mine.
(Score: 2) by Grishnakh on Friday May 02 2014, @09:13PM
No, not like that at all. The cartridges cost more than hard drives of the same capacity. You're better off just backing up on HDs for those prices. What I'm asking for is a reasonably-priced drive, and inexpensive media. $164 for a 1TB cartridge is expensive when I can get a 1TB HD for $60 or $70.
(Score: 2) by Jaruzel on Friday May 02 2014, @09:40PM
Ah... I don't think you are going to find that. I've resorted to buying one of those external USB to SATA 'docks' and I plan on getting loads of 500tb HDDs as my backup media - right now that seems the most cost effective way of backing up large data.
This is my opinion, there are many others, but this one is mine.
(Score: 2) by Jaruzel on Friday May 02 2014, @09:43PM
Oops, 500GB HDDs I mean. 500TB HDDs would solve the problem permanently!
This is my opinion, there are many others, but this one is mine.
(Score: 2) by Grishnakh on Sunday May 04 2014, @03:15PM
I think you'd do better with an internal SATA dock; the USB-to-SATA conversion slows down your data transfer rate significantly.
(Score: 2) by Jaruzel on Monday May 05 2014, @12:37PM
It's a USB3.0 dock, but I have noticed that Windows 8.1 becomes really sluggish when there's a disk plugged into it :(
I'll have a look for an internal dock see what they're like.
This is my opinion, there are many others, but this one is mine.
(Score: 2) by Grishnakh on Monday May 05 2014, @02:09PM
I got one from "Kingwin" a while ago pretty cheaply.
(Score: 2) by isostatic on Saturday May 03 2014, @10:01AM
Lot 5 tapes, 5 pack for £50 ($80), that's 7.5TB for the price of your 1TB drive.
I've got a small online storage array of about 1.2PB, it's raid 6 with about 10:2 parity, and there's an rsync to a mirror.
A real backup of that requires a tape robot with 1,000 tapes, so about $15k for tapes. That's a lot cheaper than even 100tb of on disk storage.
Lot is not designed to backup your porn collection.
(Score: 2) by TheRaven on Saturday May 03 2014, @01:13PM
sudo mod me up