"15-20 kilowatts per hour"
I thought better of you, Reg. I mean, this isn't the BBC, is it?
Shipped tape capacity passed 6.6 exabytes in the third 2014 quarter, a record. An HP backup, recovery and archive product manager, Simon Watkins, revealed this in a blog. As a man with lots of tape libraries in his set of products, you’d expect him to be pleased. The data came from the Santa Clara Consulting Group. Watkins …
"thereby saving 15-20 kilowatts per hour for each PB that’s not on disk."
A 3½" hard disk is about 10 Watts. If we assume modest 2TB disks and mirroring, that is 10 Watts per TB or 10kW per PB. Add in RAID controllers and switches and you could get to 15-20kW/PB - if you spin all the disks all the time.
Where does that "per hour" come from? I could understand 15kWh/hour/PB. If killowatthours per hour sound stupid, there is a reason. Just cancel out the hours. The Register does not have a simple unit of power, but one can be constructed by multiplying force by velocity: Norris ⨉ (Percentage of maximum velocity of sheep in a vacuum). Like many units in physics, two things multiplied together. Units do not have to be something per something else.
I felt sorry for myself during my apprenticeship in 70's, handling ICL exchangeable 30MB discs and tape reels up my arm, in the wee hours. My female colleagues wouldn't lift the discs for fear of dropping the things.
But I wouldn't fancy carrying many of those Marconi reels - each circa 25lbs?
Those were manly days :-;
Shipped capacity will keep growing, even if the market is static, surely? People will want bigger tapes to stay where they are - able to back up X user's worth of data - but as that data grows, the capacity shipped will increase.
In the same way, I could say I'm now shipping MILLIONS OF TIMES more capacity around the globe than I was in the 80's. We all could. It just means we popped an SD card in a bag and posted it, rather than a cassette tape from a ZX Spectrum.
Surely what matters in terms of the business and market is the VALUE of shipped items, not the capacity of them?
1c per gigabyte is also not that much to brag about (and the cheaper it is, the worse the above hits you, no?). That's $100 for 1Tb. You can buy hard drives at that price on Amazon, without even trying. Just because tape is catching up with the very drives they backup, it's not much to crow over.
Because they are hiding actual dollar figures behind lies, damn lies, and statistics, and because of my own experiences, I'll got for the exact opposite: "Tape is dying" Multiple NAS, even offsite, offline, etc. are starting to do the jobs that tape once did. Cloud backup is perfectly viable as one-of-many backups. The only advantage for tape is WORM capability, but with proper storage plans that can be enforced anyway.
The one thing that ALWAYS worried me whenever a serious situation occurred that might prompt me to check my tape strategy in case it was needed: How much is it going to cost if the drive that reads the tape doesn't work and I have to source a replacement? Sometimes, the answer was more than the value of the data it stored. Tape itself might be "cheap" per Gb, but the surrounding hardware, warranties and support contracts won't be.
Tape's dead. Any technology that takes longer than my users are logged off overnight to backup what they did yesterday is dead. Just make sure you have enough redundancy to cope.
I'm just a lowly amateur photographer with 10's of thousands of RAW files backed up on a little computer I built with a few 1TB drives I got for less than that tape costs... What's wrong with backing up to a hard drive, then unplugging it? They don't REALLY rust over time, do they? $0.01 per GB or $100 per TB is not even remotely close to a good deal.
1c per gigabyte is also not that much to brag about (and the cheaper it is, the worse the above hits you, no?). That's $100 for 1Tb. You can buy hard drives at that price on Amazon, without even trying. Just because tape is catching up with the very drives they backup, it's not much to crow over.
Ahem, 1024 GB in a TB, at $0.01/GB that makes $10.24...
Tape is not dead, you just have no use for it. Luckily, we're not all you.
"1c per gigabyte is also not that much to brag about (and the cheaper it is, the worse the above hits you, no?). That's $100 for 1Tb."
You might want to recalculate that figure ...
Hint - there are 100 cents to a dollar, and 1024GB to a TB
1024/100 = ???
Plus don't forget the saving of 15-20 exosheep per parsec, or whatever the figure was.
1000 x 1c is $10 not $100.
I think the rest of your comment is based on your view from your IT infrastructure rather than a real understanding of what is going on and your last comment about backups taking longer than your users are logged off mean you don't have enough of it or you haven't got it working properly.
Indeed they were classic. Made by people who manufactured ceramic sewer pipe as they were the only ones who could make a cylinder that would spin that fast without wobbling, flying apart or shaking the frame to bits.
Also, one needed to assemble an hydraulic crane to the frame to lift 'em out in the event they needed servicing.
Does data storage get any more manly?
>15-20 kilowatts per hour for each PB that’s not on disk
What is a kilowatt per hour?
Presumably they mean kilowatts or maybe killowatt hours. But if its killowatt hours then over what period?
Say 512 2TB drives are used to store 1PB and that the drives are on and spinning 24 hours a day.
Also lets assume that it takes no power to retain the 1PB on tape.
If each drive is 6 watts then the power use of the drives would be
512*6=3072 watts lets call this 3kW
So 15-20 figure seems rather high but you could use smaller more power hungry drives.
Per day these drives would use
3kW*24=72kW hours per day
In this case the 15-20 figure seems rather low. Maybe they are factoring in the option to power down the hard drives if the data is not needed for a while.
No denying there could be a significant energy saving, but it would be nice if they at least used appropriate units.
Oh and obligatory shame on the Reg for not noticing.
The thing that is killing tape isn't the tapes, its the insane cost of entry compared to disk. A small business can pick up an inexpensive NAS box with a couple 1 TB disks in it with some basic backup software for less than $1500-2000. If more storage is needed, it is relatively inexpensive to just buy a new box, often offering better features.
Compare that to tape, which, at the lowest end, will require a $5000+ tape drive, a server to connect to it, tapes, and storage for the tapes (which have a much smaller maximum write count than disk). That isn't even considering the cost of purchasing a separate system to test the tapes... And that isn't even considering license cost of tape library software.
Backup admins are, by nature, a very risk-averse group of people. If disk-to-disk backup has worked for them in the past, they are very unlikely to change no matter the cost savings. Even then, management would also be quite adverse to such a change, since the business depends on being able to recover quickly from disaster.
It's true that disk has trashed the economics of tape at the low end for on-premise, but luckily for us in the tape industry the cloud people are hoovering up all those small requirements and presenting them in PetaByte-sized chunks. The cloud is an astounding growth areas for tape...
normal measure is Total Addressable Market
6.6 exabytes = 6.6 Billion Gigabytes. But at 1c per GB, that's only £180M per annum.
Split that among a few companies worldwide, and it's a hobby not a business
Would struggle to justify more than £10M R&D across the whole tech, and what are you going to achieve with that?
A 4 TB HDD costs $120, retail qty 1. That's $.03/GB. Without a quantity discount and without compression. The LTO guys very frequently hype their numbers by assuming compression. It is a heck of a lot easier to compress disk ("de-dupe") than tape, but let's just assume HDD will be 3x LTO at the module level.
The equipment to support a a PB of HDD is a lot cheaper than the equipment to support a PB of LTO.
You can turn off the disks you aren't using, so the power for the disk system is at least a small as for the tape and probably smaller.
disks can be re-read or even reused millions of times. An LTO can be read or written a total of 260 times.
You can access any data in a PB of spun-down disks in less than 10 seconds.
Tape is dead.
Your whole post breaks down because you are not comparing apples for apples.
LTO is an enterprise grade storage system, for instance it is 1000 times less likely to suffer bit rot than your disk drive from PC world........ If you read back your entire 4GB drive you have a 1 in 2.5 chance of a file corruption. you would have to read an entire LTO-6 tape 16,000 times to have the same chance of an error. (this is a bit of cheat because the LTO is 2.5TB but you get the picture)
Entrusting business critical information on a consumer grade drive. Price a 2Tb sas enterprise quality drive certified for HP/Dell servers or any good storage array. Bit more than $100ish. At 58TB a weeknight for incrementals and over 140TB of weekly fulls, and a 7 year storage requirement, backup to dedupe with short retention times and constantly cloning a full to a 5 drive tape library is the only way I can go. That is unless they suddenly build me a new datacenter where I would need 2 aisles of disk sitting there, mostly doing nothing and most assuredly not meeting my Disaster Recovery requirements. So while watching the leds flash on my Data Domains is entertaining, watching the tape library robot run and cycle tapes beats it hands down. Especially when if I need more capacity its $50 for a tape and the library will even automatically spit out the ones I need to box up for Iron Mountain. Small operations, fine, put it on disk. Fortune 100 companies with more money than sense and a datacenter the size of a warehouse, fine put it on disk. Heck you do what you want. Its your data. Tape is not dead in my neck of the woods, not by a long shot. Thank God no one mentioned cloud backup or I would have had a stroke.
Sure you can probably make the drive compress your database dumps 4 fold. However how often do you have the case that you need to store huge amounts of readily compressible files on a machine that does not have gzip installed?
So 1 cent per Gigabyte probably means more like 4 cents per Gigabyte, which is right in the ball park of cheap harddisks. You can turn off / spin down harddisks and you don't have the huge overhead of a tape changing robot and multiple drives (plus maintainance).
The raw figures are here:
http://www.lto.org/2015/03/lto-program-announces-price-per-gigabyte-now-less-than-one-penny/
Yes: they are assuming you fit 6.25TB of compressible data onto a 2.5TB LTO-6 tape, at a cost of $50.50 for the cartridge. Ignoring the compression you are talking 2c/GB ($20/TB) for the cost of tape.
This still compares favourably with the cost of hard drives: amazon.com shows $160 for a 4TB WD Red or $120 for a 3TB WD Red, both of which are twice as much per GB as the tape cartridge.
The hard drive also needs a permanent SATA slot - either powered 24x7, or with switchable power. The LTO-6 tape needs drive(s) and a robot to load it, and a decent climate-controlled environment to store the tapes.
So as well as the price/GB, the economics involves the cost of the robot and what ratio of tapes to drives you are happy to run with.