Tape users wait for news of LTO 7 and 8
Virtual roadmap ends with blank, featureless desert
Comment According to published plans, the all-conquering LTO tape format has no future after LTO 6, which is expected in 2012. This is ridiculous and there must be a secret roadmap for LTOs 7 and 8.
The LTO tape format dominates the open systems tape business, is used increasingly by supercomputer customers and is making some inroads into the mainframe area via bridges such as those from Luminex or Fujitsu's CentricStore (ETERNUS CS) virtual tape library (VTL). Supercomputer users like Argonne and NASA Ames in the USA will have ten year roadmaps. They can generate 8PB or more of data a year and there is simply no alternative to tape for storing that amount of data.
They are currently envisioning a move to the LTO 5 format, with a 1.5TB native capacity and 140MB/sec throughput. There is currently a 3-year window between tape generations. We can expect - as the LTO Consortium has a public roadmap - an LTO 6 format with 3TB capacity and up to 20 percent improvement in throughput to 168MB/sec in 2012*. After that... nothing.
What are NASA Ames, Argonne and other supercomputer and large scale enterprise LTO users expected to do? They'll keep on growing data. We could soon be looking at double digit petabytes being generated annually by large supercomputer users - indeed, CERN's LHC project is already looking at 15PB/year.
Are they meant to stay on a 3TB LTO 6 cartridge tape format and see the number of tapes they need to cope with the increased amounts of data rocket up, while data write and read times steadily lengthen because 168MB/sec is too slow?
Are they supposed to move to disk, because future arrays with deduplicated 6TB and larger drives will be cheaper than tape? The large tape automation vendors can't see this happening, because the drive arrays will be physically huge and, unless significant amounts of spin-down takes place, the power budget will be astronomical.
It's a nonsense to them. Of course there is an LTO roadmap out past LTO 6. It's just not yet a committed roadmap for the consortium and its two main drive manufacturers, HP and IBM**. We can map this virtual roadmap out ourselves using the 3-year gap, capacity doubling and bandwidth 20 per cent increase algorithm.
LTO 7 will have 6TB native capacity, a 201MB/sec throughput and appear in 2015. LTO 8 will jump to a 12TB capacity, a 221MB/sec bandwidth and debut in 2018.
We can be fairly confident that there is broad agreement about this inside the LTO consortium. Either that, or it has a secret migration plan for LTO 6 tape automation users to move to an undisclosed alternative platform, such as a de-duping MAID array using 6 to 8TB hard drives, with the data reliability and throughput of a large-scale tape library.
Which is it? LTO 7 and 8 or the MAID array? Supercomputer and enterprise LTO tape users are getting restive and want to know what they should plan for. ®
* The published roadmap has a 270MB/sec native bandwidth for LTO 6, but bandwidth improvements seem to run at 20 per cent per generation now and not 100 per cent.
** According to a person familiar with the situation, Quantum only make a statistically insignificant number of drives.
Sponsored: Benefits from the lessons learned in HPC