The LTO Consortium members are finally looking beyond 2006's LTO 4 generation and predicting a 3TB plus cartridge with a LTO 6 format, expected around 2010/11. This puts it behind Quantum which is suggesting its new DLT-S6 format offering 3-3.5TB will appear in mid-2009, one to two years earlier. But Quantum has gone further and has proposed a DLT-S7 generation for mid-2011 with 6-8TB capacity per cartridge.

Is the LTO consortium being over-cautious?

LTO members are talking about extending the LTO roadmap by two new generations - LTO 5 and LTO 6.

LTO 3 products are just coming available. They offer 400GB native capacity - 800GB compressed - and transfer data at 40MB/s native. The already-known LTO 4 will arrive up to two years after LTO 3 and offer 800GB raw capacity and a transfer speed of 80MB/s (native), both numbers twice that of LTO 3.

LTO 5 will offer double the 800GB native capacity of LTO 4, meaning 1.6TB - that's 3.2TB compressed at 2:1 compression. It will appear about 18-24 months after LT0 4 in c2009. Speed should go up but may not double.

LTO 6 will offer twice as much capacity as LTO 5, meaning 3.2TB native and 6.4TB compressed. It will appear around 2010/11. Again speed will probably go up but not double.

It seems that Quantum has stolen a march on the LTO group by accelerating its development and delivering the same 3TB capacity cartridge up to two years earlier, and providing a migration path to a 6-8TB cartridge. Graham Hunt, Quantum's EMEA marketing manager, confirms, "Mid-2009 is 3.5TB native by Quantum (DLT-S6)."

Why is there this disparity in roadmaps? Brad Renfree, director of strategic customer management for Certance, pointed out that, "There is a leapfrog effect. LTO has a capacity advantage for a time; then Quantum. then LTO again. That doesn't bother us in the least. The market clearly likes LTO. Analysts say it has some 70 percent of the supertape market."

What about Quantum's DLT-S7 with its 6-8TB capacity for 2011? Renfree says that they think issuing a roadmap, "with the current generation plus two is sufficient for IT planning." It keeps things simple. "LTO 7, if the market asks for that, is something we would formulate when LTO 4 is ready."

Assuming that the market keeps on requiring a doubling of capacity that would produce a 6.4TB (native) LTO 7 format in 2012/13.

Renfree is not convinced a doubling of speed is needed with each generation; "If we continued doubling transfer speed then LTO 6 would be moving a gigabyte a second. It's technically feasible but not necessarily right for the market." The implication is that LTO transfer speeds will rise but by less than a factor of two.

WORM differences
The LTO group is also going to match Quantum, and IBM, Storagetek and Sony, by offering a WORM - write once; read many - tape. Certance has already discussed its LTO 3 WORM product.

LTO WORM capability is based on a different cartridge from the standard LTO catridge. The drive will recognise the different cartridge and then write to it once if it is an empty WORM tape and refuse to write to it if it is already fully written to.

Quantum's approach is to use a normal cartridge and its DLTice technology to provide WORM facilities via a firmware addition to existing SDLT 600 drives. It will be standard in new drives. The drive recognises a WORM-type header on the tape and will refuse to write over any data areas.

In both instances the drive, via firmware, responds to a WORM status signal coming directly or indirectly from the cartridge loaded into it. We might ask why having the signal on the cartridge case or on the tape inside it is significant? The LTO people suggest that having a WORM tape that is basically reusable unless a SW written-header is detected by a WORM-aware drive is fundamentally flawed.

According to Hunt, "When you deploy WORM in the real world you might not actually have to keep WORM tapes for ever. You might only need to keep it for one year." So it makes sense, he says. to be able to re-use WORM tapes that are no longer needed rather than have to bear additional media purchase costs.