Original URL: http://www.theregister.co.uk/2012/10/19/ibm_125tb_tape/
IBM prepares to demo 125TB MONSTER tape
Huge LTO-6 hardness is shingled
IP Expo: Tape Summit IBM has revealed it is preparing a technology demonstration of a 125TB tape, and has revealed that LTO-6 tapes use shingling, with overlapped data tracks.
In January 2010, IBM demonstrated a tape with 35TB of raw capacity. Apply LTO-6's 2.5:1 compression ratio to that and you get 87.5TB. This contrasted with the then-current LTO-5 tape's raw capacity of 1.5TB.
The LTO consortium has a roadmap of two more formats: LTO-7 with 6.48TB raw capacity, and LTO-8 with 12.8TB. Assuming that the upcoming LTO format capacity doubling, IBM's 35TB tape, created with the help of Fujifilm, would fit between a conceivable LTO-9 with around $25TB and an LTO-10 with 50TB.
At the time we said such a tape would take three days to fill at LTO-5 speeds. This does not matter as it is restore speed that is vital, and a fast streaming drive would get to the start point of a file needing restoration in seconds.
At the Tape Summit, Paul Scheuer, tape brand management programme director in Big Blue's Storage Systems Division, said IBM was developing a 125TB tape – 3.6 times more capacious than the 35TB tape. It is able to do this because the physical size of a bit on tape is many times larger than the size of a bit on disk, and can be shrunk without having to go to the difficult and expensive bit-patterned media and energy-assisted recording techniques facing disk drive manufacturers in their quest to increase areal density.
A tape read/write head doesn't have to move anything like as much as a disk head, which snakes in and out across a disk platter's surface, hunting for the destination data track and then following it. Scheuer said that a tape read:write head also read from narrower tracks than it wrote, and write tracks were overlapped, shingled like roofing tiles, so more of them could be put down along a tape's surface.
Write tracks on the 125TB tape are overlapped, like shingles on a roof.
By increasing the sensitivity and bit granularity of the read:write heads and shrinking the physical size of the bits, to a 100Gbits/in 2 areal density, a 125TB capacity tape using a development of today's barium ferrite recording media technology, should be attainable. This could theoretically enable an LTO-11 format with 100TB raw capacity and 250TB compressed capacity.
Scheuer pointed out that tape cartridge capacity needed to develop so as to maintain a consistent advantage over 3.5-inch disk drive capacity. We have 4TB disks today and LTO-6 tapes with 6.25TB compressed capacity. As long as successive tape formats sustain the capacity advantage over disk then tape's cost/GB advantages over disk should be sustained as well.
There is a prospect here of the LTFS file:folder tape interface combining with a 100TB or more tape cartridges to provide a highly attractive and dense storage medium for businesses needing access to lots of large files, such as high-definition videos and other large data sets that individuals need to work on. With LTFS and such tape cartridge capacities, the notion that tape could be an individual or small workgroup's backup medium could perhaps be re-evaluated. ®