Merc, any speculation on where the break-even point is for storage? Of course, the loss of on-line capabilities is tough to quantify, but at what point does buying tapes instead of drives pay for the library system?
Thing with tape is that it will cost X number of dollars for the hardware (tape drive(s), library enclosure that may fit in the rack or standalone) - (I'm putting aside backup licensing costs etc).
Tape will also cost Y in terms of media costs. Say Y = 30TB.
In comparison we have a rack of spinning disk. It will cost A for 30TB.
The point of crossover will be when your data storage needs exceed the limits of Y or A.
In the tape scenario, hardware cost X remains the same, you simply increment Y costs (in terms of getting more tapes).
In the disk scenario, hardware cost X must either be duplicated or upgraded. In most cases the tape expansion will be cheaper, especially as the TB/PB quantity increases. We aren't talking about drives from Newegg, RAID drives are premium priced by storage vendors and basically a choke point. Tapes on the other hand can be multisourced and are interchangeable (assuming similar types and that LTO goes back to read/write the previous generation and read only on the generation prior.
X+Y(amount of expansion) versus X(amount of expansion).
Like you say though, it's not one or the other. It's a combination of both. Multiple redundant arrays can provide more potential uptime in case of primary array failure, but when the Virus comes (or more likely a disgruntled employee) to deliver the plague, I can only sleep at night knowing at least a reasonably recent copy sitting on tape in an Iron Mountain vault.
BTW Ultrium 5 aka LTO 5 will be with us shortly: 1.5TB and 140MB/sec native, 3TB and 280MB/sec assuming max 2:1 compression.
http://www.webnewswire.com/node/517466
This will drop the pricing on LTO 4 which will remain the volume seller whilst LTO 5 commands premium pricing in the early adopter phase.
