bit-tech.net

Hitachi to ship world's first 1TB hard drive?

Hitachi to ship world's first 1TB hard drive?

Hitachi may well be the first to ship a 1TB hard drive.

Within hours of each other last night, both Seagate and Hitachi announced plans for their first 1TB hard drives, which will ship in the first half of this year.

However, it looks like Hitachi will be the first out of the gate with its Deskstar 7K1000 drive, as it looks set to begin shipping to retail customers during 1Q07 with a suggested retail price of US$399 ($0.40 per GB).

Seagate, on the other hand, has said that it will make its 1TB Barracuda available to the public in the first half of 2007.

The Deskstar 7K1000 will be available in both SATA 3Gb/s and ATA133 versions, with the SATA version boasting a whopping 32MB on-board cache. Meanwhile, those wanting to buy the PATA version of the drive will have to settle for only 8MB of cache.

Hitachi's new 1TB drives use five 200GB platters that spin at 7200RPM, and they're the company's first drives to make use of perpendicular recording technology. However, it will not be the first drive to make use of perpendicular recording technology, as Seagate was first to use that technology in 3.5" drives when it launched its 750GB Barracuda 7200.10 drive last year.

Hitachi has also announced plans for a four-platter 750GB drive later this year too, although retail pricing hasn't been disclosed at this time. Both of these new high-capacity drives will make use of three low-power idle modes that will boost power efficiency (and should reduce noise levels) when the drive isn't seeking.

30 Comments

Discuss in the forums Reply
DougEdey 5th January 2007, 12:10 Quote
40c per GB is good, hope it translates directly over here, 30p per GB

EDIT: Is Q1 for the UK aswell?
Mister_Tad 5th January 2007, 12:54 Quote
Makes a nice solution for my offsite backups. High capacity tapes just aren't cost effective on a small scale.

May have to pick one of these bad boys up from the states when they're available if they're only going to make a $400 dent.
Firehed 5th January 2007, 13:04 Quote
Hope the price is right, not too far off the current sweet spot. 32MB cache? *drool*
LeeV 5th January 2007, 13:12 Quote
I dont think it will have a place on the computer for the moment but I think it will be good for PVRs or as a Backup drive, but puting so much data onto one drive is bad if it fails. You should buy several cheap drives and spread your data out on them.
mikeuk2004 5th January 2007, 13:16 Quote
I dread to think how long it takes to format. Be prepared to not need to use your computer for a few days or weeks while it formats :p
Tim S 5th January 2007, 13:18 Quote
Quote:
Originally Posted by LeeV
I dont think it will have a place on the computer for the moment but I think it will be good for PVRs or as a Backup drive, but puting so much data onto one drive is bad if it fails. You should buy several cheap drives and spread your data out on them.
There's a DVR/Set top box version too, known as CinemaStar.

Welcome to the forums. :)
JazX101 5th January 2007, 14:17 Quote
Very nice, space enough to swing many cats, or whatever other animal you think might be disk compatable??
One thing that puzzles me is the way that hard disk technology is only advancing in data density: we have had 7200 spin disks as standard for years (not being entirely sure on how many, figures anyone??) without any notable change to the setup. Yes we now have SATA II, which increases the transfer speed while you're empyting the buffer, which takes only fractions of a second, for sustained transfer you're not even maxing out the P-ATA standard. Along with NTFS disk managment and NCQ with some drives, you're getting a slight improvement in performance. The only real change from standard is the Raptor drives, with their 10K spindle speed mean you can suck the data off faster, but how many people have the money to buy them?? You could also buy 15K spin disks, if you had a spare kidney you wernt using and wanted to sell.
Why dont the manufacturers actually improve the technology behind the drives themselves?? :
1Gb of flash ram is now easily under £20, if it were intigrated into the controller board of the drive, you could have the boot files for Windows on there, slashing boot-up times to daft levels! You could even have a flash ram drive, running programs only, giving split-second start up times to any application, then having a seperate Hard disk for data, files and media??
Increasing standard spindle speeds would be a bit more of a challange, but it would be possible if the interest was there from the market!
Something slightly more radical: If the hard disk case were to be extended for another couple of centimeters, there would be enough space for a second stack of read-write heads, allowing data to be taken off a disk at twice the speed, written at twice the speed, halving search times etc. You could even watch media from one part of the disk while writing new data to another part (think PC-TiVo without any hanging or lost scenes) The only area that would need development for that would be the control circuitry, to combine the 2 data streams into a single data stream to the computer, and to split the incoming tasks into single head tasks. Just think of the speeds on a dual head drive that also made use of higher spindle speeds, and large buffer/flash storage areas? mmmmm
Anyhow, enough of my coffee fueled ranting, I'm going to go watch a film!
Jaz_knos
Mister_Tad 5th January 2007, 14:41 Quote
The only manufacturer other than WD that there is a chance of seeing 10k or 15k SATA drives from is Samsung, as they (like WD) have no enterprise SCSI/SAS disks to protect.

low access times present little to no benefit to single-user access patterns so there isnt a great deal of need for them in the consumer market. Most enthusiasts aren't prepared to pay the premium on raptors given the slight real-world benefit seen from them, so the average consumer certainly isn't ready for it.
Quote:
Something slightly more radical: If the hard disk case were to be extended for another couple of centimeters, there would be enough space for a second stack of read-write heads, allowing data to be taken off a disk at twice the speed, written at twice the speed, halving search times etc.

it isnt radical at all, its called RAID0, and it presents very little benefit (if any at all) in a desktop environment. This would also double the complexity of the drive, double the heat in the drive, likely more than double the cost and double the chance of failure.
Sord_Fish 5th January 2007, 14:44 Quote
Woo hoo just what i've been waiting for. Will be getting one of these for a diy NAS box hopefully ext3 wont be too bad on this :D

wheres winfs when you need it hehe.
DXR_13KE 5th January 2007, 16:52 Quote
once again: HUGE DRIVE FOR PORN!!!!!!!!! :D lets see if this doesn't cost a leg.
Mister_Tad 5th January 2007, 16:57 Quote
Did you read the article? $400
JazX101 5th January 2007, 18:27 Quote
Most of the heat in a drive comes from the central spindle, not the drive heads. I know that for the majority of the time, most people will never see their hard drive, or any other component for that matter, used to its maximum. Surely as members of the enthusiast community, who can get most twitched about minute differences in benchmarks, its not a matter of "well most people dont care" more a matter of "what if?"
Jaz_knos
Mister_Tad 5th January 2007, 18:39 Quote
The hard drive industry's bleeding edge, however, is not driven by the enthusiast community.
HourBeforeDawn 5th January 2007, 23:22 Quote
hmm Im going to wait until Seagate has theres out to offer a possible price drop but cool I have been waiting for a 1TB drive for my HTPC project ^^
Lowsidex2 6th January 2007, 04:32 Quote
Bah... I'll get excited when they release a drive with 1TB of formatted space. What will this have, less that 900gig usable? False advertising IMO.
David_Fitzy 6th January 2007, 05:20 Quote
Quote:
Quote:
Something slightly more radical: If the hard disk case were to be extended for another couple of centimeters, there would be enough space for a second stack of read-write heads, allowing data to be taken off a disk at twice the speed, written at twice the speed, halving search times etc.
it isnt radical at all, its called RAID0, and it presents very little benefit (if any at all) in a desktop environment. This would also double the complexity of the drive, double the heat in the drive, likely more than double the cost and double the chance of failure.
I reckon JazX101 has come up with the same something I dreamt up a while ago. I wouldn't call it raid 0 as you're using the same surface to read from. Raid means redundancy the only redundancy would be if one head failed the second could still be used.
I think he means two sets of heads per stack of platters (one front, one back) coupled with an updated NCQ I reckon it'd work really well doubling the throughput(ish) even going upto 3 maybe 4 sets of heads (in a wider package)
Like this:
http://www.fitzpatricks.plus.com/David/drive.jpg
A new NCQ could not only reorder packets/requests to be written/read efficiently it could assign them to a head depending on the head's position on the drive. Idling the heads could sit at the 1/4 mark and 3/4 mark to improve access times, heavy contigous access could use both heads at the same position reading alternate tracks.
Cthippo 6th January 2007, 05:29 Quote
Quote:
Originally Posted by David_Fitzy
I reckon JazX101 has come up with the same something I dreamt up a while ago. I wouldn't call it raid 0 as you're using the same surface to read from. Raid means redundancy the only redundancy would be if one head failed the second could still be used.
I think he means two sets of heads per stack of platters (one front, one back) coupled with an updated NQC I reckon it'd work really well doubling the throughput(ish) even going upto 3 maybe 4 sets of heads (in a wider package)
Like this:
http://www.fitzpatricks.plus.com/David/drive.jpg

Great idea! The SATA and especially SAS systems can already move more data than the drives can squirt at them, so doubling the read capacity might finally allow drives to live up to the potential of their busses.
David_Fitzy 6th January 2007, 06:45 Quote
To go to the extreme...

http://www.fitzpatricks.plus.com/David/drive4.jpg

Someone forward this to R&D at seagate (all i want is a freebie (or 8) when they're made)

EDIT: Imagine putting a drive window on this
r4tch3t 6th January 2007, 07:35 Quote
I think quite a few of us have thought of this before (I was actually thinking one for Read, one for Write, but this design does that and can also get faster reads and writes.)
The only problem would be more advanced firmware to orient the heads to work together. And if it was retrieving a large file and was using both heads, the OS would most likely have to piece it together.
As for the RAID 0, does it actually get different parts of the file? I thought both Hard drives retrieved the whole file.
David_Fitzy 6th January 2007, 08:03 Quote
Quote:
Originally Posted by r4tch3t
I think quite a few of us have thought of this before (I was actually thinking one for Read, one for Write, but this design does that and can also get faster reads and writes.)
I don't deny that it's a fairly logical step
Quote:

The only problem would be more advanced firmware to orient the heads to work together. And if it was retrieving a large file and was using both heads, the OS would most likely have to piece it together.
That's been half solved by NCQ, it just needs to queue commands into 2 (or more) stacks based on the positions of the heads
Quote:

As for the RAID 0, does it actually get different parts of the file? I thought both Hard drives retrieved the whole file.
I think there might have been some confusion over what the raid levels mean, I'm allways confusing raid 0 and raid 1 My favourite site for raid levels
Raid 0 (Striping) splits files into blocks and distributes them among 2 or more drives. This improves Read and write times at the expense of reliability (If one drive fails the data on the other is useless. Technically it isn't RAID as it doesn't offer redundancy)
Raid 1 (Mirror) writes the same files to both drives and reads the same files from both drives. This doesn't effect read/write times
Raid 1.5 (note not 15 or 1+5) reads alternate blocks from each hard drive speeding read processes but not write (in theory, reading alternate blocks requires seek time tomshardware on raid 1.5)
Mister_Tad 6th January 2007, 08:07 Quote
r4tch3t: only a part of a given file is on each drive with RAID0, ie in a 4 disk array, 256k of a 1MB file will be on each drive.

I think most of us have probably thought of multiple heads in a drive before. The main problem with this though: there isn't any market for it.
Cthippo 6th January 2007, 08:24 Quote
Quote:
Originally Posted by Mister_Tad
I think most of us have probably thought of multiple heads in a drive before. The main problem with this though: there isn't any market for it.

What do you base that conclusion on? I guess the real underlying question is where is the bottleneck for hard drives? It seems to me that the issue is getting the data off the platters is the problem that hasn't been solved despite the years that drives have been raound. Capacities have gotten larger, spindle speeds have increased, but as I understand it the read technology hasn't changed that much. Has anyone ever tried a multi-haded drive? Seems like it would work very well for the server market where each head could handle a seperate request and so 4 heads could handle 4 different requests.

Sounds like a questin for Tim.
Mister_Tad 6th January 2007, 08:38 Quote
Quote:
Originally Posted by Cthippo
Seems like it would work very well for the server market where each head could handle a seperate request and so 4 heads could handle 4 different requests.
Various RAID levels have populated servers since the beginning of time (well, nearly) doing just this.
Quote:
Originally Posted by Cthippo
I guess the real underlying question is where is the bottleneck for hard drives?
I work with disk arrays with several hundreds of disks, several GB of bandwidth and hundreds of thousands of IOPS, no bottlenecks here.
Bottlenecks only tend to exist in systems where hard disk performance isn't a priority - home systems.

I'm not sure that Tim's guru'dom of hard disks is quite up to the standard of most other areas of technology, though I'd be interested if he's got any input.
Cthippo 6th January 2007, 09:16 Quote
Ok, maybe bottleneck is the wrong term. What is the limiting factor on drive performance at the current state of tech? Or, is it your view that current hard drive tech is entirely adequate to the needs of existing platforms?

Personally, I'd like to see standard drive capabilities move towards those of the i-Ram without the volatility.

Maybe it's just me, but I feel ike there is a good deal of room for improvement over my 10k RPM SCSI drive.
Mister_Tad 6th January 2007, 09:49 Quote
Quote:
Originally Posted by Cthippo
Ok, maybe bottleneck is the wrong term. What is the limiting factor on drive performance at the current state of tech?
The limting factor on current hard drive tech is that there are moving parts.
Quote:
Or, is it your view that current hard drive tech is entirely adequate to the needs of existing platforms?

in my view? absolutely not! I'd love to see some cost-effective solid-state performance drives become available, but until then I'm stuck with high RPM drives for the OS and apps and RAID5 for mass storage.

For enterprise use, current hard drive tech is adequate because of the scalability factor - one drive not fast enough? get a hundred. 100 drives not fast enough? then get a thousand, and so on.

For the vast majority of hard-drive buying public, a hard drive isnt a performance piece of equipment like a CPU or grapics card - Bigger is better, not faster.
Quote:

Personally, I'd like to see standard drive capabilities move towards those of the i-Ram without the volatility.
You can purchase solid state non-volatile drives, but they are prohibitively expensive. A mechanical hard drive will never be able to reach the performance level of a solid-state drive. Hybrid drives are to address this. Given the relative phenomenal cost per GB of solid state storage not only is it not feasible to make a 500GB solid state drive but its also unnecessary. Who needs to access their pr0n at 50000 iops? (other than Bindi, I mean). Easy solution, a few GB of onboard solid state storage for performance sensitive data and a traditional platters + heads solution for mass storage.
Quote:

Maybe it's just me, but I feel ike there is a good deal of room for improvement over my 10k RPM SCSI drive.
A bit off topic, but what drives are they? SCSI drives have never really been optimised for a desktop system. Even current 15k drives are surpassed on the desktop by the most recent WD Raptor (as a matter of fact i replaced my 15K MAU with a raptor as my system drive just last week)
IccleD 8th January 2007, 15:53 Quote
Quote:
Originally Posted by Lowsidex2
Bah... I'll get excited when they release a drive with 1TB of formatted space. What will this have, less that 900gig usable? False advertising IMO.

I agree, 1TB on the Box, means 1000Gb, which we all (should) know isn't a True Terabyte. After formatting your looking at even less. Using http://www.wiebetech.com/whitepapers/BillionEqualBillion.pdf
We can work out the following: (Hard Drive Manufacturers use Decimal Maths!)

A decimal TeraByte is 1,000,000,000,000 which is equivalent to:
1,000 x 1,000 x 1,000 x 1,000 = 1,000,000,000,000(Bytes) = 1 TB.

A binary TeraByte is calculated in the following fashion:
1,024 x 1,024 x 1,024 x 1,024 = 1,099,511,627,776(Bytes) = 1 TB.

Thats almost a 10% Difference in Capacity,
1,000,000,000,000 / 1024 = 976,562,500 (Kilobytes)
976,562,500 / 1024 = 953,674.316 (Megabytes)
953,674.316 / 1024 = 931.322574 (Gigabytes)

So the "TeraByte" Drives being released are infact only 931.3Gb in size, thats 93Gb short of the true Terabyte figure. Hell, thats more than the total space of some new Laptop Hard Drives.

The use of Decimal over Binary on the Box by the manufacturers leads to a misleading of the general public, and a false statement of storage capacity.

I'm just gonna stick with my 3x 400Gb Drives, in a Raid, which not only give a True Terabyte of storage, but is cheaper (just) than one of these new drives anyway!

FYI, thats 3x400gb = 3x380 True Size @ £85 each (Dabs WD400) = £225 for 1200Gb = 1,140Gb. Thats over a Terabyte, with my figures rounded down also.

Considering that a current 500Gb drive from Seagate is about £150-£200, and even two of these is under a Terabyte, I can't see a 1TB drive for less than £300, which still makes my Raid better value for money.
Renoir 8th January 2007, 16:25 Quote
You can get 500GB drives for under £115 now Samsung Sata 500GB
JazX101 8th January 2007, 19:05 Quote
Thanks for the Diagram, Thought i would make it a bit more fun:
http://img479.imageshack.us/img479/4408/dualheadharddrivecl6.jpg

There would be benefits to this setup, beyond the enthusiast community. one of the main reasons against would be the deviation in the standard hard disk enclosure size, and its compatability with cases, racks, servers etc
Jaz_knos
Mister_Tad 8th January 2007, 19:53 Quote
Quote:
Originally Posted by JazX101
There would be benefits to this setup, beyond the enthusiast community

A) Can I ask, on what are you basing this? which brings me to point... B) this thread has gone a bit off topic, if anyone wishes to further discuss HDD design please do so in a new thread in the HW forum.
JazX101 9th January 2007, 02:11 Quote
Server usage, higher data flow, symultanious read/write. IMO Adding a second set of drive heads shouldnt cost the world, and for a marginal increase in price for a notable increase in performance it would make sense. Besides, within a few years we will have all the storage space we could want (are you going to fill a multi terabyte drive any time soon?? i've only just managed too fill my 300Gb one) improving the performance of a drive will become the next competition area.

I've started a Therortical Hardware! thread in Hardware and Overclocking for anyone who wishes to continue with the thought train, or offer any ideas they have had for new tech. I hope for some interesting responces.

The link again for those who didnt get the blueness the first time around:
Therortical Hardware!

Jaz_knos
Log in

You are not logged in, please login with your forum account below. If you don't already have an account please register to start contributing.



Discuss in the forums