Therein lies a big part of the problem, why should their be a "side". Science is about the finding of fact and facts don't care what side you're on. You might want to deny that a certain type of virus won't affect you because of your religious belief for example. Problem is, the virus doesn't care, don't have a "side" and will kill you just as well as everyone on the other "side". People can deny all kinds of things as much as they like, but in the final measure, it doesn't matter squat, the climate will change, you will get lung cancer, HPV will infect you, whatever...

Cost is subject to volume it's also relative. Consider just how overbudget things like the UK's Trident nuclear weapon system is or the JSF for that matter.
It wasn't so long ago that an energy transmission cable was proposed from Iceland to the UK and Continental Europe so that all that geothermal energy could feed the mainland beast. Having flown over that part of the world a fiar bit, let me tell you Iceland is a one hell of lot further than Scotland's minor islands.

Speaking as an Enterprise vendor: 1. Nobody ewver pays list. 2. You aren't just paying for the device, there's a tonne of development going on. One dedupe appliance I worked with had over 120 engineers behind it that all have to be paid, plus every time you put something into the market, you find all kinds of weird-ass coner-cases that have to be diagnosed, debugged and fixed. You have global manufacturing, logistics and 24x7 support infrastructures to pay for. We also try and engineer-in more reliability, redundancy and durability than you can buy off-theshelf. LTO, for example, is *two to three orders of magnitude more reliable* than consumer hard disks.

Fine, YOU ship an EMC VNX for $300 and then we can talk.

As LTO can compress (for free) and encrypt (usually licensed) in hardware, I'd rather hope that your NetApp compression (and deduplication) is also free or are you paying for that? Even there you get better reduction ratios for less money if you purchased a dedupe appliance from Quantum, HP or EMC.
Tape is seldom the bottleneck if you have sized it correctly. If the tape is running slow (e.g. an LTO-6 drive running at 60MB/s) then it is the disk array that cannot supply it data fast enough. If the drive is running at 160MB/s then it's maxxing out (assuming you get no compression which today assumes 320MB/s). The vast majority of business arrays are optimised for IOPS and backup is a _sustained_sequential_ workload and once you empty their cache's most arrays just can't keep tape drive buffers stuffed, but few storage admin have the testicular fortitude to admit their big-$ array can't do sustained sequential workloads very well.

One more reason SSD's are such an improvement is they seldom have trouble keeping tape streaming and thus make tape work far better than disk ever could.

Finally, at the hundreds of TB, or in the peta-scale, disk is simply unsustainable at volume, between purchase, licensing, support contracts, power and cooling and generation migrations every 3 to 5 years.

Your home NAS is not the same problem organisations with very, *very* large datasets have to solve and thus very different cost-structures.

As LTO can compress (for free) and encrypt (usually licensed) in hardware, I'd rather hope that your NetApp compression (and deduplication) is also free or are you paying for that? Even there you get better reduction ratios for less money if you purchased a dedupe appliance from Quantum, HP or EMC.
Tape is seldom the bottleneck if you have sized it correctly. If the tape is running slow (e.g. an LTO-6 drive running at 60MB/s) then it is the disk array that cannot supply it data fast enough. If the drive is running at 160MB/s then it's maxxing out (assuming you get no compression). The vast majority of business arrays are optimised for IOPS and backup is a *sequential* workload and once you empty their cache's most arrays just can't keep tape drive buffers stuffed, but few storage admin have the testicular fortitude to admit their big-$ array can't do sustained sequential workloads very well.

One reason SSD's are so good is they never have trouble keeping tape streaming and thus make tape far more reliable than disk.

Finally, at the hundreds of TB, or in the peta-scale, disk is simply unsustainable at volume, between purchase, licensing, support contracts, power and cooling and generation migrations every 3 to 5 years.

Your home NAS and I daresay (what sounds like) your single NetApp NAS, are not the same problems organisations with very, very large datasets have to solve and thus very different cost-structures.

To get a drop-in replacement for an existing tape library so that you don't have to rebuild your entire backup workflow overnight.

If you look at most deduping PBBA's like Quantum's DXi range or HP's D2D, you can see they allow you to emulate a tape library as a *non-disruptive* drop-in replacement and they also let you creat SMb or NFS targets too so as new backup sets are created or as old tape sets expire out of rotation, new backup jobs can be created on the LAN instead.

Don't forget - what works _for_you_ may not work for the hundreds of thousands of other businesses worldwide.

I used to work at an LTO manufacturer and asked why we never drove the older generations down into the SMB space and it is simply this - the components are *really* expensive, the majority of the component cost of the drive is the R/W head, that alone probably accounts for 25% of the drive and you just can't push the price down much further, it costs what it costs. Also, the HUGE majority of these things go into libraries with hundreds of drives, thousands of slots and robots that can move upwards of 90km per hour.

Several years ago when the overall tape market was declining, this was essentially due to the growth of LTO being masked as it cannibalised all the other tape formats (DAT, DLT, SAIT, et al), the overall number of LTO media shipments has continued to increase, that is, PB's shipped.

Two tape-centric factors are in play; capacities keep getting massively bigger but there are fewer customers that can actually use up all the available capacity. Spooks, arguably, but there are lot fewer intelligence agencies in the world than the small and medium-sized businesses that make up the bulk (around 80 percent) of the global economy. The Entertainment industries sure like LTO, its capaciousness and reliability has proven ideal for archiving the digital masters of their SD/HD/4k/IMAX/onwards and upwards formats. Though again, not that many when compared to the global economy.

The second factor is, everyone's known LTO-7 has been coming for a while and tape purchasing cycles always slow down around the introduction of a new capacity point. Organisations usually skip a generation (people who bought LTO-3, probably skipped four and upgraded to five) and once they do buy a new generation, usually buy a smaller library as they can now store double the capacity in a library half the size (and cost).

Like any tech, once the easy science and engineering is done, the market shakes out and the few reamining players begin to consolidate, usually down to one or two as tape has done, as disk is now doing and as SSD's will do in the next couple of years. Right now the only companies doing fundamental physics and materials research into tape are IBM and Fujifilm. Quantum no longer makes its own drives, HP will not make its own LTO-7, leaving everybody buying off IBM while the long-tail business windows down. IBM has played the same game here thay played with mainframes, they doubled-down and invested in new technology when everyone else was giving up in the face of Windows and PC's. The mainframe busines is still a $2bn per annum business and will remain a significant chunk o' change for many years to come. (Arguably, it's actually growing in some places...) That's a nice business model where all the costs have been sunk and what's left is maintenance margin. Well-played IBM. (As long as IBM's tape business can survive the sinking revenues of it's disk business which it's lumped in with).

Maybe to survive LTO will roll into a proper joint-venture, single manufacturer, where HP, IBM, Quantum and perhaps Oracle, throw in their IP to keep the drive technology best-of-breed and keep their share of that long-tail business. (Don't hold you breath though, too many ego's in that equation). Maybe it'll spin out into a niche business like OpenVMS has.

Given the problems the disk manufacturers appear to be having in shipping their new tech (SMD and HAMR) to the public in volume and the rise of SSD's, given that there is no significant amount of disk in (the massive global) archive, it's likely hard disks will die off well before tape does as it's far easier to swap out todays primary arrays for SSD's than it will be to migrate the mass of archives on tape.

Immigration? Seriously? Puh-lease, go cry to the Native Americans already. How about those "annexed" Hawaiians who then had their land filled up with "immigrants" from the USA until a large enough number of them had moved in to vote for statehood. You worried that's what South Americans might do to your little paradise too? Turnabout's a bitch. Suck it up.

Massive? Well, that's relative. Last time I checked the plan it was most certainly NOT the plan to have a collection of small daemons long-term, many more functions will be rolled into it over time.

Pilkington's "Activ" glass will self-clean in the rain. Though with credit to your comment, they recommend hosing it during prolonged dry spells.