Points off Mr Smith. You've forgotten to use the universal disk size comparison scale; How many songs will this enable you to put onto an iPod?

Paris wouldn't make such a school boy error, even on a Friday

What were the figures, a 13nm hole can produce 1Tb/sq inch.

This machine makes 17nm holes... Which means about 0.6 Tb/sq Inch.

Current hard drives generally have 25Gb/sq", although some as high as 65Gb.

So this is on an order of a tenscale improvement over current tech. Which with 3 platter drives provides us with 1TB drives at the moment.

So we can expect 10 TB drives once this comes on stream.

It's not a quantum leap, but it is more than the standard incremental improvements we see continously.

Imagine the datalosses possible with this technology. On Deskstar GXP.

//Svein

And for those kinds of storage capacities it's lucky that ZFS is around:

http://breden.org.uk/2008/03/02/a-home-fileserver-using-zfs/

a life times worth of porn on one disk, drool

"it is more than the standard incremental improvements we see continuously"

Love to be a pedant, but hard disk space is growing exponentially, not incrementally :-)

Mine's the one I stole from Moore/Kryder

This tech on IBM / Hitachi DeathStars doesn't bear thinking about! It would be total suicide :)

1PB...

but that will actually be 1000TB, that will actually be 1000GB... meh

If you read about it there will be up to 1200 disks per hour / per system on the Sindre HDD that differs from the standard Sindre Nano Massproduction systems.

A New Era of Computer Storage has begun.

May the Bits be with you :D

Quantum means small.

I'd of thought here, of all places, people would get that right.

PS I was very restrained there and didn't employ Mr Caps Lock to make my point, although I suspect he should have been involved.

Mark Lockwood: "Points off Mr Smith. You've forgotten to use the universal disk size comparison scale; How many songs will this enable you to put onto an iPod?"

When I was a lad, the "universal disk size comparison scale" was the telephone book.

Storage capacity for songs, on the other hand, was measured in "C's" (C60, C90 or C120).

<Gets zimmer>

So that'll be space enough and then some for my Cher and Mariah Carey collection.

More data on a single disk?

As the man said it,

It's time to put all your eggs in one basket,

Dare to be stupid!

Al Jankovic link: http://youtube.com/watch?v=nll8-kSlq6c

@David Corbett: That's an excellent metric - Reg please specify new storage tech as having a "song capacity of nnn C90 cassettes"!

That's a bit slow. Even the 1200 mentioned in a previous comment doesn't sound like much for a full production system.

If you measure disk capacity relative to it's porn storage capacity, then the unit would presumably be a Jerk?

(Paris, obviously and most relevently)

Another good question, is how fast can these holes be read? If the disk is spinning at normal 7200 rpm, then we could see a nice bump in hdd performance... likewise, we could see lower power/slower spinning for mobile devices. Or it could be incredibly hard to read those holes, and things are slow.

...as they always have been my favourite mobile hdd maker.

And to all those that say it's a problem to put a lot of data on one drive... How is 1TB not a lot of data to lose in one failure? Even 250GB can be completely priceless, which is why you run a raid if you have large discs. 3 10TB drives in raid5 are still better than 3 1TB drives in raid5. Segmenting your data to make sure you only lose a fraction of it in the event of a hdd failure is one of the more nonsensical methods that I've heard of.

I'd **have** thought that here, of all places, someone calling out an elementary error would not then make one of their own.

Three roundtrips to the moon, as measured in punched paper tape. Now THAT'S progress!

They'll have to count them all.

As Nexox Enigma said, just use a RAID. If all your data fits on one disk, RAID 1 maximizes redundancy and minimizes access time.

About 15 to 20 Terabyte in 2.5" form factor. Rising to maybe 40 Terabyte Raw storage...

They may use 50% for code redundancy to do error correction giving maybe 7 Terabyte at start and more than 20 Terabyte later.

How many centuries of MP3 listening?

I vote the ElReg unit of storage is years of MP3s or YOMs

Yup, you got me there... I hope my mum isn't reading this (she's an English teacher)

RAID 1 does nothing for access times, and unless you have a good controller can kill your write times. If you've got the spare cash for 2 mirrored drives, drop the size of each drive in half and go RAID 5... since usable space is (N-1) * S. Either way, use an appropriate strip size for the files you'll be storing.

Can't wait for this, I'm up to my first 1tb now, and I don't to spend another £115 for a backup, let alone £115 for extra space!

I'm waiting for drive space to drop before I really go hidef, because I dislike being chained to bunches of little plastic bits. 10tb per drive and now we're talking about reasonable numbers of bluray movies or recorded HD shows. Yay!

Also, I find it amusing how somebody always pops up to whine that it's impossible to use all that space and there's no good reason and you're just going to lose it all anyway... the same story every time with 100mb, 250mb, 500mb, 1gb, 10gb, 40gb, 100gb, 250gb, 500gb...

am i the only one to notice that the demonstration image looks a lot like it was produced by a greek builder - with various mis-sized magnetic dots and several un-etched areas

if thats the best they could manage for a demonstration image it doesnt really inspire confidence in the long term reliability of the media...

hx

Yes, I too remember measuring storage in feet - the number of feet of equivalent paper tape that a CNC machine tool would have to equal the storage of a bubble memory card.

My wife and I used to torment our IT friends by discussing the various bargain disk drives at the Egghead outlet store in terms of price per foot of capacity. Nearly as mind-bending as the El Reg standards of measure.

Mine's the one with hydraulic fluid stains on the cuffs and metal shavings on the collar...

As people said here before, if you want to worry less about losing large amounts of data, use redundancy in the form of extra disks to store parity data. Then you need something to control it all.

Hardware RAID used to be the choice, but costs were high for NVRAM to protect against data loss due to incomplete stripes being written during power loss. But as RAID systems use a proprietary format, you were trusting all your data to that vendor. Good luck! The argument for hardware RAID was due to (1) guarding against data loss due to incomplete stripes (power loss), and (2) as processor speeds were slow, the stripe calculations used a large percentage of processor power, so hardware RAID used a dedicated controller to take the load off the main processor.

Luckily that's all changed. Processors are cheap now and speeds are incredible, resulting in 90-something percent idle processors. ZFS says let's use some of that spare processor capacity and put it to good use for stripe/parity calculations. ZFS has the advantage of needing no extra hardware, unlike previous hardware RAID controllers. Also, as ZFS is software, free and open source, the format is not a proprietary black box. ZFS obviates the need for NVRAM by using copy-on-write and transactions, therefore avoiding inconsistent disk state due to power loss. When you create your storage array you can specify to have single-parity or double-parity to allow your storage to survive one or two drives failing, plus you can specify multiple hot spares to be used when drives fail. Using these levels of redundancy & hot spares, plus doing proper backups makes data loss extremely unlikely, even when you accept that drives WILL fail at some time.

Once ZFS, you never go back :)

I've written-up a load of stuff about ZFS and how to use it, and I've provided a load of links to Sun docs if you wish to discover more about this revolutionary file system that will make hardware RAID history:

http://breden.org.uk/2008/03/02/a-home-fileserver-using-zfs/

The storage game has been cranked up a few notches and competitors are looking over their shoulder at ZFS -- just take a look at Net Apps.

And no, I don't work for Sun, I'm just a happy user of ZFS ;-)

to install the next generation Microsoft O/S then.

Mine's the Apple White one.

So these are going to be for servers? Or are they cutting 4in wafers into 3.5in ones.

Forgive my ignorance, but can this technology be used to up the capacity of flash media. Now that would be exciting...

Actually, quantum doesn't mean small, it means the smallest discrete unit...

Quantum Bigfoots weren't...

Paris, because data storage is getting dense.

"Love to be a pedant, but hard disk space is growing exponentially, not incrementally :-)"

An increment doesn't have to be defined as an absolute, so the growth can be exponential and still be described as incremental. The increment is just defined as a function of the current size.

"Quantum means small."

Definition of "Quantum Leap" which is the phrase I used.

"In physics, a quantum leap or quantum jump is a change of an electron from one energy state to another within an atom. It is discontinuous; the electron jumps from one energy level to another instantaneously"

A discontinuous change, hence the common phrase "A quantum leap" meaning (according to princeton) "a sudden large increase or advance;"

If you're going to be a pedant you need to practice more.

how many of these nano holes would fill the Albert Hall

.....gets his Sgt Peppers psychodelic coat

Those machines are the Massproductionmachines for Optical components, Optical media and for displays and such. What they missed in the article is the Sindre HDD and the Sindre 60 HDD that are specially made for producing HDD's

Those have up to 1200 substrates per hour. (Or actuallt the Sindre HDD, The Sindre 60 HDD is semiautomatic and for developing)

Check the www.obducat.com homepage. They really have great stuff for the future.

Beam me up scotty :D