New Life for Magnetic TapeA boost in storage density could keep tapes spinning for years.
Music lovers may have long forsaken them, but magnetic tapes still reign supreme when it comes to storing vast amounts of digital data. And new research from IBM and Fujifilm could ensure that tape remains the mass storage medium of choice for years to come for at least a decade.
At IBM's Zurich Research Laboratories in Switzerland, researchers have developed a new tape material and a novel tape-reading technology. In combination, they can store 29.5 billion bits per square inch, which translates to a cartridge capable of holding around 35 terabytes of data--more than 40 times the capacity of cartridges currently available, and several times more than a hard disk of comparable size. The researchers used a relatively new magnetic medium, called barium ferrite. In cooperation with researchers from Fujifilm's labs in Japan, they orientated the barium ferrite magnetic particles so that their magnetic fields protrude perpendicularly from the tape, instead of lengthways. This means that more bits can be crammed into a given area, and the magnetic fields are stronger. Furthermore, these particles allow thinner tape to be used, meaning12 percent more tape can be stored on a single spooled cartridge. Increasing the density of data that can be stored on a tape makes it more difficult to reliably read information. This is already a problem because of electromagnetic interference and because the heads themselves will retain a certain amount of residual magnetism from readings. To overcome this, the IBM group developed new signal processing algorithms that simultaneously process data and predict the effect that electromagnetic noise will have on subsequent readings. Hard disks can store more data on a given surface area than magnetic tape, and the data on a disk can be read faster. But because hundreds of meters of tape can be spooled on a single cartridge, the overall volumetric data density of tape is higher, says Evangelos Eleftheriou, head of the Storage Technologies group at IBM Zurich. Crucially, tape storage is also much cheaper. "What's most important is the cost per gigabyte," says Eleftheriou. Solid state drives cost between $3 and $20 per gigabyte. In contrast, it costs less than a cent per gigabyte to store information on magnetic tape. In the third quarter of 2009, the global tape market was worth more than half a billion dollars. Extending the life of magnetic tape technology could delay the arrival of new storage technologies, particularly holographic storage. Experimental holographic discs, which use patterns of light interference to hold multiple pieces of data at a single point, can already hold several hundred gigabytes of data. The technology is expected to eventually allow terabytes of data to be held on a disc. "Tape still wins, but only at very high data volumes," says James Hamilton, a vice president and distinguished engineer on Amazon's Web services team, in Bellevue, WA. Tape is most suitable for "cold storage"--when data is not accessed frequently. But the volume of digital data that needs to be stored is increasing rapidly, so Hamilton says there's a real need to try to squeeze more out of tape. It could take another five years before the new tape technology is ready for the market, Eleftheriou admits. "But we have shown that there is still at least another 10 years of life in it," he says. |









Comments
There are probably 2 reasons: (1) the tape gets flexed every time the tape is read, which damages the oxide layer a little bit every time, (2) when the tape is sitting in storage, each piece of tape is in close contact with regions having different magnetization patterns.
But whatever the reason, the #1 requirement of a storage medium isn't density, it's RELIABILITY. For any data that matters, tape is a non-starter.
Neil2009
01/25/2010
Posts:3
But I would argue that tape is a necessary part of any storage strategy (and will probably remain so for many years to come). Disk replication is a great solution for disaster recovery, but can't (at a realistic cost) hold the multiple copies needed for archival recovery (i.e. when you need to return to the version of a file from 3 days/weeks ago).
Other solutions based on CD/DVD technology don't currently match the speed of tape - and many organizations struggle to fit their backup activities into the narrowing timeslot available.
chrisjmiller
01/25/2010
Posts:37
There have been several reports floating around the web about the data loss phenomenon with these media and I have experienced it first hand a while ago when I transferred 100 CD/DVD backups to HD of only a few years old (mean=2years). About 30% were completely unreadable and in about 20% parts of the data were corrupted. So I don't want to imagine the data loss you'd get when a holo-disc fails, especially considering that more complex systems are generally less robust...
I think the conclusion is that at this time we do not have a digital medium that can compare to the longevity achieved by old media like books and stone tablets... There is of course little incentive for companies to develop such a thing, but if we want to store our digital stuff for generations to come we need such a thing. Otherwise our civilisation will suffer from a general kind of amnesia, "forgetting" everything from 30 years ago.
nanobody
01/25/2010
Posts:3
Over the last decade, tape reliability has improved 700%--important, because some of that old tape was a real nightmare. (DLT comes to mind, and AIT.) However, LTO-4 and enterprise drive technology such as the IBM jag drives, with the proper care and feeding, are so far the only viable (and yes, reliable) method of storing data for anything approaching the useful lifetime of most data. Accessing data from tape that's been stored for a long time does assume that the right drives and software to read the tape are available. Otherwise, migrating data up through tape generations, which is standard in very large data centers, is a another step to take. And LTO, for example, reads and writes back one generation, and reads two generations back, so having a few LTO drive generations in a single contemporary automated tape library keeps data plenty accessible.
bethwalk
02/26/2010
Posts:1
mbelvadi
01/25/2010
Posts:15
On the positive side, a small number of bit errors in an image file are unlikely to be a visible problem, whereas the same is not the case for a file of (say) bank account data.
chrisjmiller
01/25/2010
Posts:37
I have to agree about tape. Its cheap, fast and easy. Not many a business would turn that down as a backup option no matter how short the life of tape is.
matayp
01/25/2010
Posts:2
wrampart
01/25/2010
Posts:1
matayp
01/25/2010
Posts:2
>then what happens to your data if there is a disaster at the offline backup place? what if it is only in one location and flood/fire/earthquake/theft/ degradation of storage media then happens at the online backup? (lightning strike after flood caves in roof for example) or theft of data centers computers as happend at major data center in chicago
as the previous poster says, the 'bits still have to come to rest somewhere' and that somewhere still can have reliability problems. just that somewere won't have the same unreliability problems you will 'on site' as you say.
"You also have the benefit of always on connections, your data streams constantly providing you with the most up to date data at the backup service."
>> what if disaster such as satellites going out of service (war, solar flare) or fiber trunk cut (dork with a backhoe) wipes out connectivity when you need it to the backup site?
Yes your bits have come to rest elsewhere but from this perspective how is that a bad thing?
>> this CAN be a bad thing if the bits are not WITH YOU and can't be reached when you need them. You still need to plan redundancy rather than just presuming 'the cloud' is redundant and backed up. This is similar to telecom redundancy planning. you actually ask the multiple carriers about physical routes, locations, etc.
erbium
01/27/2010
Posts:143
sactinko
01/27/2010
Posts:1
mbelvadi
01/28/2010
Posts:15