Slashdot Mirror
Seagate Announces 750GB Hard Drives
Hack Jandy writes "Seagate documents have leaked out the two 750GB 7200.10 Barracuda hard drives. The drives are the first desktop hard drives to use perpendicular recording, feature a 16MB cache and 7200RPM spindle."
Check out the Seagate Barracuda for more info.
You don't even post a link to it.
We can finally Get Perpendicular!"
DYWYPI?
I suppose that why it has perpendicular recording.
We all know 16k of storage is more than enough for anyone.
Now I have to go to Costco and buy 3 250GB drives!
This issue is a bit more complicated than you think.
that will hold almost half of my porn!
I am not left-handed, either!
Each time the capacity of hard drives goes up a few gigs, I think back to the day in the mid 90's when I got my first "gig" hard drive for $500. Wow, it was the most incredible thing to be one of the first people in my neighborhood to have so much storage... I didn't think I'd ever run out of that much space. And today, the OS won't even fit into such a thing.
But let's put this huge capacity into perspective: Having once had to reverse engineer an obsolete 3.5" floppy drive to repair an obsolete piece of industrial machinery that was down (the customer couldn't afford to replace the whole machine because of a failed floppy drive, and the OS loads from floppy of all things), I learned that this contraption, which was on the market in the 80's, was really incredible, if you take a step back and think about it for a minute. Then, all it takes is a moment to realize that hard disk drives are several orders of magnitude more complex. First, the density of a floppy drive is nothing compared to that of a hard disk even from a decade ago, and secondly, the linear motion of the reading head on a floppy is controlled by a simple stepper motor, whereas the round motion of the reading heads on a hard drive is controlled by servo. I mean, just stop to think about it for a moment. All those gigs of MP3s, videos, and pr0n on someone's hard drive, and what an incredible piece of engineering behind them.
Everyones using USB disks for backups now rather than tapes. So many benefits there. Thats why Lacie and Maxtor are making a killing on selling drive + MCU + USB + casing packages. How many small and medium sized companies have total data exceeding 750GB?
Even more interesting is who will release the first terabyte drive and (this is what I'm interested in) who will be the first to put one terabyte on a single platter. A terabyte is a lot. It will be a lot 5 years later, and quite a lot even 10 years later. Sure I understand Moores law and how 10MB was huge back then. But there comes a time after which we actually run out of relevant data to put on it. Pictures will go upto 10 megapixels but it will stop there. Video might go upto 1024x768x32-bitx100FPS but will not exceed that. Our humans senses will cease to notice any further difference. Games might require 2 blue-ray DVDs but will not require say 32 blue-ray DVDs in the next 10 years. What will you PUT on it?
Maybe this will mean I'll finally have as much space in hotmail as I have in gmail.
"Give orange me give eat orange me eat orange give me eat orange give me you." -Nim Chimpsky
Because I've experienced data loss before. That's a lot of valuable stuff (at least in my case) in a very small space with little to back it up with except for more of the same. It scares the bejesus out of me.
But I remember saying that about them huge 9GB drives when they came out when I was 12 (or so.)
750 (hard drive manufacturer GB) = 698.49 (real GB or GiB, depending on how anal you are).
As these sizes keep getting bigger the need to settle on one method of calculating GB, for both OSes and hard drive manufacturers, keeps getting painfully clearer.
Come on, there is no way that a 7,200RPM drive will have an average latency of 4.16ms, that's the pure physical latency of the platter! The transfer rate is similarly bogus, it's the burst transfer rate of the interface, not even the outer track transfer speed. Guess we have to wait for someone like storagereview to throw iometer at this beast and get some real info.
There are 4 boxes to use in the defense of liberty: soap, ballot, jury, ammo. Use in that order. Starting now.
sounds like somebody needs some porn ;)
They're becoming IO-bound far faster than cache-bound. It takes literally hours to read an entire 500gb hard drive at this point. The cache, on the other hand, is staying roughly on par with the IO speed, which seems like a more natural combination.
Breaking Into the Industry - A development log about starting a game studio.
Video might go upto 1024x768x32-bitx100FPS but will not exceed that
Right. Tell that to any gamer running @ 1280x1024. Higher resolutions will always be in demand. Games will continue to have better and better textures, more units, bigger and more maps. I wouldn't be supprised to see 1TB games in the next 10 years.
You make a good point, but just don't put finite limits on things which are likely to change quickly.
Obligatory Soundbite Catchphrase
It's getting to the point where you want to keep your OS and core applications in Flash memory and things that are less important on hard drives. I just bought a 512 MB usb key for $25. Scaling up, you could get a multi-GB flash drive for a couple hundred bucks.
Some companies have multi-tiered storage solutions (e.g. fast SCSI RAID, cheap EIDE RAID, optical, etc.). Some of those ideas may make their way into desktop devices. You'd boot off of flash memory nearly instantly (it would cache your OS and core applications), then you'd play your MP3s, surf the web, or whatever on your relatively slow hard drive.
Will Linux support it... hehehe that's cute.
...
I have friends who have multi-TB raids at their homes using a mix of IDE/Sata/USB in one RAID
While hardware RAID support in Linux is a bit hit or miss the software kernel support works properly and is fairly quick. Certainly the bottleneck for most setups will always be the drives themselves.
Tom
Someday, I'll have a real sig.
Yes, it should considering that according to Wikipedia http://en.wikipedia.org/wiki/NTFS [wikipedia.org] states that the maximum volume size for an NTFS volume is 16EiB. One exibyte is 1,152,921,504,606,846,976 bytes, so 16 exibytes = 18,446,744,073,709,551,616 bytes. Since a 750GB hard drive should hold approximately 750,170,112,000 bytes, an NTFS volume should be able to handle 24,590,081 of those 750GB hard drives in a RAID array. Now assuming a RAID array can handle that many of these drives, and that this new 750GB hard drive merely takes the price spot of Seagate's current finest offering of a 500GB hard drive (priced on newegg as $295 each) http://www.newegg.com/Product/Product.asp?Item=N82 E16822148108 [newegg.com] rather than debuting at a higher price point, which it probably will, that many hard drives would cost about $6,147,520,250 before tax, and not including any of the massive discounts one might expect to recieve for such a massive purchase. On top of that, at a sales tax rate of 7.75%, the tax on those drives would cost you $476,432,819.38. So I don't know about you, but I doubt this is going to be a problem for either XP or Vista for a long, long time (assuming you use NTFS partitions).
A quick look to Wikipedia says...32TiB for the largest volume size. If you are using a partitioning tool - that might be your limitation but it is definatly not in the file system. http://en.wikipedia.org/wiki/Ext3
That ain't happening for a while, even with perpendicular recording.
If you check out the datasheet for the 7200.10 series Barracudas (PDF), on page 2 you'll see a row with the heading "Heads/discs".
I'm going to take a wild guess and say that "discs" refers to the number of platters in the drive. Also, Seagate has the option of writing to one or both sides of the platter, which helps explain how the 200 & 250GB models have 3 Heads and only 2 discs.
So: The 750GB model will have 8 read/write heads and 4 platters, meaning they're cramming roughly 190GB per platter. IIRC, the IBM 75GB Deathstars had 5 platters instead of 4, which contributed greatly to their failure rate, so Seagate is doing the smart thing and trying to increase the GB/platter instead of the GB/drive. They're awfully close to a terabyte drive... if they used 5 platters.
[Fuck Beta]
o0t!
the 750 GB hd is really only about 700 GB due to the manufacturers counting 1,000 instead of 1,024..
Anyway, lets look at how much space that really is, and how easy it is to fill up.
DVD Movies range from 4gb to 9gb depending on film length and extras, lets settle on an easy middle number, 7GB average.
That is around 100 DVD's you could store on your hard drive (My room mate owns over 150 DVDs, so while it might be a large number to some, it is not so large to others)
That is not including TV series, if someone were to store 1 season of the show 24 on their media center pc it would take 45GB of space.
Also concider that HD movies are going to be around 30GB each
Video games are getting increasingly large, Recent games like
The Godfather (4.5gb installed)
LOTR: Battle for Middle Earth II (5gb installed)
TES: Oblivion (6.3gb installed)
World of Warcraft (5.3gb installed)
Tomg Raider: Legends ( 7.3gb installed)
Games are only going to get larger too.
This is not even counting people who dabble with video editing or anything like that, work-wise that consumes monsterous ammounts of HD space..
Yeah, then it gets cracked like 2 weeks later. I think the RIAA would be quite pissed.
The soylentnews experiment has been a dismal failure.
While your post remind people that different definitions of GB are used, you are actually adding to the existing confusion. Because what you call a "real GB" is not real at all. You should rather call it "conventional GB", as in "conventional Giga prefix used in computing, ie 2**30". The real Giga is the Giga prefix as defined by SI, ie 10**9. Disk manufacturers are just using the standard Giga SI prefix instead of the "conventional Giga prefix". Other people are doing it in the computing industry. Bandwidth and throughput are also typically referred to using standard SI prefixes (e.g. an MP3 file at 128 Kbit/s is 128000 bit/s NOT 128*1024 bit/s).
The filesystem may reserve enough bits to address 16EiB, but that doesn't mean Windows can handle it yet.
The idea here is that you might be able to make a monolithic head using MOS techniques and cheap-up the manufacturing process. The surface area of a 2.5" HDD is so small, we're not talking about a huge acreage of silicon. And the magnetic coils you would need are just little round circuits, aren't they? I'd have thought that would be amenable to some form of photo etch fab process. If you had to go 3D you might even use a fabber (3D printer, see http://www.ennex.com/ ) for much of it.
Do not mock my vision of impractical footwear
An important point seems to be missed by everyone in all the "1Tb won't run out in a few years", "yes it will", "no it won't" discussions. Given more space, engineers will think of new applications for all that space.
It's not like you were filling up that 20Mb harddrive with text files.
It's not like you were filling up that 1Gb harddrive with black and white bitmaps and low fidelity samples.
And you're not going to fill that 1Tb harddrive with JPGs, movies and MP3.
3D environments (for games or other purposes) will take more and more space, as objects and their textures get more detailed. And that's just an application that's already here. Think of what you can do with all that space and think of something new.
How about CGI-movies with dozens of selectable camera angles? How about we send you all the feeds of a sports event with a direction script and let you mess with it? I'm sure you can do better than I am, just saying there -will- be new ideas. Wilder and more storage hungry than what I'm proposing here and we -will- be needing Pb drives in 10 years.
Pound the table all you want, but it simply isn't "just the way it is". Keep in mind that the http://www.essex1.com/people/speer/large.html predate computers by decades or centuries (depending on your precise definition of "computer"). According to the metric system:
The only way you could say that 1 kilobyte is 1,024 bytes is to make a special exception to the metric system's prefix rules, and the whole point of the metric system is to have a system of measurement without silly exceptions like that. If they had wanted a system where you had to memorize different rules for different units, they would have stuck with the imperial system.
So to sum up: some computer geeks thought it would be convenient for them to redefine the metric system to work using powers of two rather than powers of ten. This was fine as long as they were only interacting with other computer geeks. When computers spilled over into the world at large, however, this little shortcut conflicted with the way the terms were/are used by everyone else. Since the traditional (powers of ten) definition has both seniority and wider usage, it is now winning out, and rightly so.
I don't care if it's 90,000 hectares. That lake was not my doing.
I'm running 3x 400GB SATA's in a software RAID5 on a linux server no problem. The key problem you may be thinking of was the LBA48 support. Older IDE drive controllers only supported LBA32, so they would only see 137GB of larger drives. Often a BIOS upgrade fixes that, and ups support to LBA48. Newer SATA spec doesn't require LBA48, but I'm not aware of one that doesn't, though there's probably the odd one. There'd very likely be a BIOS or flash upgrade for the controller to add 48-bit addressing. LBA48 tops out at about 2000GB per drive on 32-bit processors IIRC.
Remember kids, it's all fun and games until someone commits wholesale galactic genocide.
I admit it. My brain went immediately to "Get Perpendicular!" ... AND THE HITACHI BRANDING ... as soon as I saw the reference to perpendicular in the story. It's been months since I saw that animated short -- if not more than a year.
Someone needs to give the guys who thought that up, a bonus.
The author of the article mis-interpreted Seagate's latency figure. Seagate means: "Average rotational latency". This can be calculated from: 60 seconds/minute / 7200 RPM / 2 = 0.00416 s = 4.16ms.
Oficially you should add in the controller overhead, and most likely the time to read a sector (it's unlikely they pass-through the sector: in theory you can start to send the sector to the host before you've read it completely, but this complicates things as when the CRC doesn't match, you have to cancel the data sent to the host!), but if you do the math, these are negligable compared to the 4.16 ms.
I don't expect anything "special" to happen in the "seek times" area. They will be within 10% from the slightly older drives. Either up to 10% better because they did find a way to improve seek times a bit. Or up to 10% worse because the higher density requires a longer settling time, but this is less likely than a small improvement.
Even if you could fabricate the head assembly, you still have a major problem. Modern track densities require closed-loop head positioning. If you could shrink yourself to the size of the head gap, you would see the head constantly moving laterally to keep itself positioned over the track. At this scale, the platter is no longer an ideal rigid disk.
Mea navis aericumbens anguillis abundat
Only a few PC owners seem to think about backups and then only after loosing their data. Even with a 1PB (not a typo) disk there will be people who won't backup. Now in business if you don't backup they have a very good chance of going bankrupt if they have a hard disk failure. It must be rembered that even mirrored or SAN disks can and do fail.
....etc, so disk performance is not normally an issue, however when you are running a database you need to use RAID or a SAN to increase the speed of the (now virtual) disk. So 10 x 72GB disks in a raid array will outperform 1 x 720GB disk.
.... etc) are used and even then tapes can fail for a variety of reasons. What is really needed is a replacement for tape and currently only the Holographic Versatile Disk (HVD) has the potential to do this with 300GB/disks to be available this year. Forget about CD, DVD, HD-DVD and Bluray these technologies have their own niche and will most likely work well within it.
For home use a huge disk is mainly used for pawn, movies, games
Backing up a business system is expensive and time consuming so currently tapes (eg: super DLT's
Even with the potential of 1.6TB/disk (40MB/sec write) for HVD the increase in storage will still cause issues in the future until people learn to tidy up their messes and that I don't see happening anytime soon. The larger the available storage the more it will be filled up with what is most likely junk, although my idea of junk (or goodness) will most likely differ from others.
There ain't no such thing as proprietary standards only proprietary formats. Standards are by definition open.
Oolite: Elite-like game. For Mac, Linux and Windows
Sorry, I can't get overly excited about a hard drive maker increasing storage space. That is all they have been doing for the last 10 years, certainly hard drive performance hasn't been driving the industry.
Hard drives are the single biggest bottleneck on today's systems. With multi-core technology and cheap gigabytes of ram all with gigabyte transfer rates, a hard drive plodding along with a 100 - 200mb/s transfer just doesn't cut it. Why should my system seem to hang with only 10% CPU utilization because of intense hard drive activity. I can't even bring up another task that doesn't use the hard drive because the system is too busy with hard drive transfers.
Either a new I/O standard needs to be invented, something that doesn't tax your system when excessive hard drive transfers are made, or the frigging hard drives just need to start getting up to gigabyte transfer rates.
In any case, I could care less about hard drives doubling or tripling in size, until they show significant improvements in performance, or move to solid state, then I am apathetic about the whole industry.
I haven't thought of anything clever to put here, but then again most of you haven't either.
Outstanding.
Doesn't have anything really to do with latency, but I've seen several comments from folks who worship at the altar of rotational speed when the true factors that determine a hard drive's speed are aa combination of rotational speed, track-to-track latency and data density. You can spin an old 10mb drive at 200,000 rpm and it still won't transfer data faster than a modern hard drive.
As sector density increases so does data throughput for a given rotational speed. If all other things are equal when you double the sector per track density you *almost* double the drive's throughput. I say almost because in order to double throughput you'd have to cut seek times in half as well.
But - fast drives have dense platters, not just fast spindles.
we see things not as as they are, but as we are.
-- anais nin
Los Alamos National Labs uses 3Com SATA raid cards in their scientific compute clusters. The 9500 series is excellent, and very well supported in Linux. I've built several multi-terrabyte cluster heads for them. Sixteen 250GB drives in one 3U case, on two 8 channel SATA raid cards doing raid 5.
Yes, I did wear ear protection while configuring and testing the things.
- None can love freedom heartily, but good men; the rest love not freedom, but license. -- John Milton
Its called SCSI. Quit buying shitty storage and then complaining that it's shitty.