Data Center Advisors

Trains, Tapes and Roadmaps

I’m writing this blog while heading to the Narita Airport from Yokohama on the brand new Narita Express. This new train is beautiful, shiny and clean with 180 degree rotating seats. All the things you expect from brand new equipment. Though I thought the old Narita Express train wasn’t too bad either. In fact, except for the new everything, the ride feels the same, it takes about the same amount of time to get to the airport and the per car capacity is about the same. There is some enhanced security with the new personal luggage locks, but not a game changer. I’m sure there are some incremental improvements here and there as well as the updated equipment, but there’s no WOW FACTOR! This got me thinking about tape roadmaps. Yes, I’m a bit punchy as I head home from a week of excellent meetings with colleagues on projects, and dining and drinking in Japan.

Products based on the LTO5 specification are due to hit shelves around March this year. This is a 1.5TB and 3TB of tape storage uncompressed and compressed (2:1) in a single tape cartridge with about maximum transfer speed of 280 MB/s. This is about twice the capacity of the current LTO4 technology which can store 800GB of data uncompressed and 1.6TB with 2:1 compression, and with a maximum transfer rate of 240MB/s. The LTO6 specs, the last committed specification at this time on the LTO public roadmap, specification is to be 3.2TB native and 6.4TB compressed with a maximum transfer rate of 540 MB/s. This is a healthy doubling of capacity about every 2 years. This is an OK and predictable pace and allows for some consistent ho-hum planning. However, it doesn’t track very well with several industry reports projecting the worldwide data growth expectations. Some studies project that there will be 73ZB – yes, zeta bytes or 73,000 exa bytes – of data to be store by 2020. Using IDC’s projections where 21% of all storage will be tape, a value I call the “Tape Factor”, this puts the share of data to be stored on tape at about 15ZB. So, I’m looking for something big to happen.

Fujifilm and IBM announced last week that they have developed and demonstrated a technology for storing as much as 35TB of data on a single LTO sized cartridge, native!. This new technology, called Nanocubic technology, is still in its early stages, but it probably won’t be ready for production for another 6-7 years. Maybe this is why the LTO7 and LTO8 specifications are still uncommitted. This is 44 times the current capacity of LTO4 tape cartridges. Now we’re talking WOW! Also, assuming the length of tape doesn’t dramatically increase and the tape moves across the tape head at about the same speed as today’s LTO4 (2m per second), that’s a potential for almost a 100GB/s read transfer rate. It will probably be slower than this but you can imagine the possibilities. Now that’s WOW-WOW and game changing.

So while I am on the “Tape is Dead for Backup” bandwagon, I believe the future is bright for tape for long term archiving of inactive data. Maybe during the next train refresh, the Narita Express will be a bullet train.