SoylentNews is people
SoylentNews
from the as-opposed-to-non-linear-tape? dept.
The Linear Tape-Open market is stable:
The LTO Program Technology Provider Companies (TPCs)—Hewlett Packard Enterprise, IBM and Quantum—today released their annual tape media shipment report, detailing quarterly and year-over-year shipments.
The report shows a record 96,000 petabytes (PB) of total compressed tape capacity shipped in 2016, an increase of 26.1 percent over the previous year. Greater LTO-7 tape technology density as well as the continuous growth in LTO-6 tape technology shipments were key contributors to this increase.
[...] While the total compressed tape capacity grew dramatically in 2016, the total volume of tape cartridges shipped in 2016 remained flat over the previous year whereas hard disk drives (HDD) saw a decrease in unit sales of approximately 9.5 percent year-over-year2. This stability in tape cartridge shipments indicates that customers continue to rely on low-cost, high-density tape as part of their current data protection and retention strategies and evolving tape technologies are becoming attractive to new areas of the market.
"Compressed tape capacity" is a nonsense number that multiplies the "raw" capacity by a compression ratio. Assuming that only LTO-6 and LTO-7 tapes were sold (which have a 2.5:1 compression ratio rather than the 2:1 of earlier generations), then 38,400 PB or 38.4 exabytes were shipped.
LTO-6 tapes store 2.5 TB and LTO-7 tapes store 6 TB. Planned LTO-8 tapes will store 12.8 TB, LTO-9 will store 26 TB, and LTO-10 will store 48 TB. The max uncompressed speed of these generations will be 160, 300, 427, 708, and 1100 MB/s respectively.
(Score: 3, Insightful) by cmn32480 on Wednesday April 19 2017, @06:29PM (11 children)
As long as you have a decent backup window, tape is convenient and cheap. Particularly in the SMB market, where cost plays a much bigger role than in larger enterprise.
Autoloaders are relatively cheap if your backups span multiple tapes, they (generally) last YEARS in a decently climate controlled environment, unlike a hard drive, which can be completely hosed by dropping it.
They can take shipping or hand carry to offsite facilities easily.
Much like terminal emulation, tape isn't going to die anytime soon.
"It's a dog eat dog world, and I'm wearing Milkbone underwear" - Norm Peterson
(Score: 5, Informative) by ledow on Wednesday April 19 2017, @07:04PM (10 children)
£2500 + £30 a 6Gb tape for a basic LTO-6 tape drive (Just googled for a brand I can recognise, that wasn't the cheapest, no autoloader etc.) buys you an awful lot of other things.
I imagine small shops aren't even bothering at that kind of price. Price is the primary reason I hear people steer clear of tape for backup, even in schools and offices.
Include autoloaders and rackmounts and you get to £5000-6000 easy.
Tape quickly becomes quite expensive, per Terabyte. And you're still looking at being down for the day to restore from a tape even if everything works.
Then you have to consider warranties, repairs, costs of sourcing a replacement in a disaster, backup sites, etc.
A lot of places are choosing not to do tape now. And tapes aren't as resilient as you make out. Sure, it might survive a fall, but it won't survive any kind of climate change for long, especially rapid change, and it needs a piece of kit that also can't survive that to read it back off.
Tape has uses, don't get me wrong (offsite and transport, as you point out, but again - you have to have another drive at the other end or move that too, and that's just going to break if you drop it, the same as a hard drive), but if it was a choice between putting £10k on a tape setup for, what, a few dozen TB in total of backups (including daily, weekly, monthly, etc.) - there are much cheaper solutions, with much faster restore times, using much more standard kit, that you can plug into little self-contained portable restore units that are everything you need to copy data back to a network in a little box. Hell, many of them can be mounted via iSCSI and just carry on as you were.
You can say "it's not as resilient" but the drive has to figure in that too, or payment for someone else to supply a drive on short notice, or costing in multiple drives, etc.
And if you can get several copies of the network onto cheaper devices, with better MTBF ratings, and in a format that itself can survive failures, and be read off using a standard drive interface or Ethernet, is that a better or worse backup? There's no one right answer to that. But for a lot of people, it works out much better not to use tape.
Honestly, the *technology* used is irrelevant, so long as you have enough backups of varying types, ages and locations. When there's anything other than tape, it's ALWAYS my last choice to restore from, however.
I've even recommended to tiny one-server shops who have asked me to help, and that are religiously doing their tape swapping: continue. But I'm going to stick this USB hard drive on here and leave it plugged in. For £50, getting "another backup" that restored in less than half the time of their tapes was a no-brainer for them. And if you have the software and the alert infrastructure, it's no harder than just adding it into the backup job as a copy. Sure, you wouldn't want to rely on JUST that, but alongside other methods it's another +1 for them. And where I've done that, every restore I've ever done has come from that drive. And they can take it off-site, and replace it just as easily as a tape when it goes wrong.
The place I work for now sacked the previous IT guy for failing to tape backups. The servers failed, the network went down, the data was lost, game over. Literally into the "pay to recover a damaged RAID array" territory, and it cost them a lot.
They hired me. But they didn't blink at not paying for tape. All the similar places in their industry were doing it. All their auditors picked no fault with it. It was just an expense that they put into different backup solutions and resiliency instead. Remove the dependency on any one technology, device, server, array, drive, etc. - even PERSON - all the way from top to bottom. Tape doesn't need to figure in that to make it robust.
Last time we were audited, the bought-in expert questioned me for hours on behalf of the insurance company (who were insuring against data loss and cyber attack, etc.). I made grand claims (restore times, number of copies, ability to restore, off-site capabilities) etc. They were shocked. HOW many copies? How frequently? So if I lose a file, you can pull it back in how quick a time? And how many previous revisions? Made how frequently? Going back how far? From how many independent backups? Literally, they were taken aback. So we did it for them. Here you go. There's the backup catalog (also stored in four separate locations, and not even necessary to restore from, it just makes it more convenient). There's the list of everything backed up. Pick something you see on the network. Anything at all. Any server, any storage, any file. Right, there it is in the backups. There are all the copies, backups, replicas, etc. with all the backup history. Pick any one. Okay, restore. There you go, it's back. Pick a VM. Any server you like. There you go, there are its checkpoints and the number of backup devices they are on. Pick one. Let's restore it. Churn, churn, churn. Working VM copy on test system.
Tape can't really do that. Not to the same extent. You always need something in combination with tape to do that. And when that's the case, the tape is merely incidental.
(Score: 2) by Grishnakh on Wednesday April 19 2017, @08:46PM (6 children)
Um, this really doesn't make sense. There *are* no cheaper backup solutions, in $/GB, than LTO tape that I'm aware of. You can get portable hard drives pretty cheaply these days with multi-TB capacities, but they're still more expensive per TB than LTO cartridges. You're right that hard-drive based online backups have massive advantages in being able to do continuous snapshots and being able to easily access those backups at any time, but it comes at a cost. If you're backing up really massive amounts of data, requiring hundreds or thousands of cartridges, LTO is the cheapest way to do it; the expense of the drive is a one-time cost, so unless you're only backing up a handful of tapes worth of data, that cost becomes insignificant. So a small business probably has no business bothering with tape, but larger organizations can do backups far more cheaply that way.
(Score: 2) by requerdanos on Wednesday April 19 2017, @10:04PM (2 children)
Um, this really doesn't make sense. There *are* no cheaper backup solutions, in $/GB, than LTO tape that I'm aware of. You can get portable hard drives pretty cheaply these days with multi-TB capacities, but they're still more expensive per TB than LTO cartridges.
I think that the point was that LTO tape is only cheaper per gigabyte if the tape drives are free, which they are not [newegg.com], and one tape is drive is needed at any site where any backup, verfication, or restore can take place.
Every hard drive includes not only its magnetic medium, but also the drive needed to read and write that medium anywhere it might be located. Not so with an LTO cartridge.
(Score: 2) by Grishnakh on Thursday April 20 2017, @05:10AM
Right, so if you're backing up hundreds of tapes' worth of data, the cost of that drive isn't very significant any more. If you're only backing up a couple tapes' worth, then it definitely is.
(Score: 1) by fyngyrz on Sunday April 23 2017, @07:48PM
The only way you'd get me to use tape backups is if (a) the drives were not expensive, (b) tapes were not expensive, (c) the drive was an awesome reel-to-reel thing that made my office look like a 20th century mad scientist's lair. (c2) I can haz blinkenlights too? (c3) I can haz Switches?
Otherwise, it's multi-terabyte drives for me. Immediate random-read-access is just the cherry on top.
I'm not a fan of video and not handling very large amounts of data, so the choice remains open to me.
Also... just referring back a few posts... seems like there's no actual need for tape drives to be that fast on write, as one could multi-stage the backup through fast temp storage. Get it off the system to a fast drive, then back up that drive to tape at something less than ludicrous speed, no?
(Score: 2) by butthurt on Thursday April 20 2017, @01:14AM (1 child)
If you're backing up really massive amounts of data, requiring hundreds or thousands of cartridges, LTO is the cheapest way to do it [...]
The previous poster seemed to be describing a place where a single hard drive could hold te whole of their data. Hard drives are now available with 12 TB uncompressed capacity; that ought to be enough for some.
https://www.hgst.com/products/hard-drives/ultrastar-he12 [hgst.com]
(Score: 2) by ledow on Thursday April 20 2017, @07:36AM
Tape has use cases.
But how many people are using hundreds of tapes? Not that many compared to how many people can back up everything to a handful of them.
For every massive company with hundreds of terabytes of vital data, there are hundreds of companies with only terabytes of vital data. Probably that same company's lawyers, accountants, architects, HR outsourcers, etc. etc. etc.
(Score: 2) by TheRaven on Thursday April 20 2017, @08:18AM
I did the maths in a previous story about tapes, so I can't be bothered to do it again now, but the conclusion was that, eventually as your storage size increases, LTO converges on being the cheapest technology. Tape has much higher fixed costs, but lower per-GB costs than other technologies. I think last time I worked it out, the break-even point for tapes vs hard disks was 100TB of backup, which is a lot more than many SMEs need. The problem for tape is that this number was about 10TB when I did the same calculation a couple of years earlier. The amount of data that most businesses have is growing a lot more slowly than the rate at which the break-even point for tape is receding[1], so the potential market is also shrinking. This pushes up the price of the drives, because each new generation needs its R&D costs amortised over fewer sales (and costs more to for the R&D in total, because tape stopped being easy decades ago), which pushes the break-even point even further out. It won't be long before the break-even point is so high that there aren't enough people left to cover the R&D costs for the new tape.
[1] This is also a problem for Oracle. 20-30 years ago, any company with more than a couple of dozen employees needed a high-end database server to manage all of their internal payroll, HR, and so on data. Now, a similar company might have 100 times as much data per employee, but a cheap commodity PC running Postgres (or even MS SQL Server) will happily handle it, so the market for people who need Oracle keeps shrinking.
sudo mod me up
(Score: 1) by Roger Murdock on Thursday April 20 2017, @01:16AM (2 children)
And you're still looking at being down for the day to restore from a tape even if everything works.
Wait what? Are you using audio cassettes?
(Score: 3, Informative) by ledow on Thursday April 20 2017, @07:32AM (1 child)
A single 6TB tape at the speeds of the tape drive I found (which states 576 GBph uncompressed transfer rate)?
That's 12 hours.
If you're talking autoloaders, you could be down for DAYS.
(Score: 1) by Roger Murdock on Friday April 21 2017, @03:30AM
Fair enough, I'm imagining that having to recover 6+ TB of data probably means there was a pretty significant storage failure though which might render snapshots/checkpoints moot. We use disk backups, volume shadow copy etc for ease of restore for "I lost my word document" moments which is probably 90% + of restores but anything older than a month or two comes from tape. Anything that requires a restore from off-site media involving multiple tapes probably also involves new hardware, which is a much bigger bottleneck than tape speed.