Huge 10k RPM+ harddrives

Posted:
in Future Apple Hardware edited January 2014
Why are we only seeing small harddrives spinning around at 10, and 15k RPM? The SCSI-disks tops out at ~74GB unformatted, and the SATA Raptor one at 36GB for now, all at 10000 RPM.



Is there some physical law that makes it impossible to make harddrives spinning that fast when using higher-density platters, is it just just unfeasible or costly to do, or is it possible that some harddrive-maker gets the balls to do it in the future?



Such disks should be very good for video-editing/streaming, I think...
«1

Comments

  • Reply 1 of 24
    Quote:

    Originally posted by Zapchud

    Why are we only seeing small harddrives spinning around at 10, and 15k RPM? The SCSI-disks tops out at ~74GB unformatted, and the SATA Raptor one at 36GB for now, all at 10000 RPM.



    Is there some physical law that makes it impossible to make harddrives spinning that fast when using higher-density platters, is it just just unfeasible or costly to do, or is it possible that some harddrive-maker gets the balls to do it in the future?



    Such disks should be very good for video-editing/streaming, I think...




    Actually, that's not quite true: IBM have a Ultra320 146GB unit that spins at 10K, and doubtless once it's market-proven and the 290GB unit is out (which will be in around 18 months if history is a guide), someone will be shipping a 15K variation.



    The thing to remember is that increased capacity does not necessarily increase performance across all parameters: seek times actually seem to degrade as the density increases.
  • Reply 2 of 24
    Just think though, (4) 18 GB 15Krpm in Raid 0. Would be really really really fast and good for Video, games, everything. Much faster than a 72GB 15K. I heard somewhere that smaller drives are faster because of the computer not having to search a large area for information. is this right.
  • Reply 3 of 24
    Quote:

    Originally posted by Altivec_2.0

    Just think though, (4) 18 GB 15Krpm in Raid 0. Would be really really really fast and good for Video, games, everything. Much faster than a 72GB 15K. I heard somewhere that smaller drives are faster because of the computer not having to search a large area for information. is this right.



    As far as seeking goes, that is correct. But, the higher capacity that the drive is, the faster it can pull off data. Because the information is closer togetherr on the platter so it can read and write a ton more information at the same speed.
  • Reply 4 of 24
    When are we going to finally dump these old ass spinning drives and finally move to solid state storage?
  • Reply 5 of 24
    Quote:

    Originally posted by the cool gut

    When are we going to finally dump these old ass spinning drives and finally move to solid state storage?



    "When the cost per gigabyte is even vaguely on a par" is the obvious answer followed by "when the density (i.e. mb/cubic centimetre) is comparable".



    People stick with what they know - "nobody got fired for buying IBM" - and so until a major industry player comes up with an alternative that addresses the needs of corporate players (i.e. the guys who pay less, but buy more, and in effect pay for the R&D of people like HP, IBM, Intel and the like), the micro-businesses and home users that are Apple archetypal market will have to rely on "stone age" spinning disks.



    This theory actually holds true in many areas, especially in the whole civil war that exists between Itanium and P4/Xeon where Intel cannot hope to sell Itanium to businesses in "critical mass" quantities whilst Xeon systems are so much more cost competitive and have the binary compatibility with P4.



    This problem is then exacerbated by the fact that Intel cannot make a "consumer move" to 64-bit until the business market has paid for the R&D of Itanic - so the only option is to deliberately remove Xeon from the business market and create a "forced migration" scenario (a bitter pill which some OS9 users have swallowed, unlike myself who moved to OS X cos it seemed like the smart thing to do).



    But if they withdraw Xeon, and Itanic's IA-32 emulation still stinks, they run the risk of driving manufacturers and businesses into the arms of AMD (whose 64-bit offerings I've read have great IA-32 performance) or - even worse - end customers - who can see better price performance - into the arms of Apple if it chooses to expand its server offerings.



    I know some of this is a bit off-topic, but it does answer your question as to how innovation gets held back as much by commercial necessity as by consumer resistance.
  • Reply 6 of 24
    dobbydobby Posts: 797member
    We use 8Gb SSD's on our solaris machines that res pics for us. They are lovely to mirror as a reync takes under a minute.



    I've go a couple of old ones on a Dual 533 with OSX and an adaptec SCSI card. The newer ones are all FCAL.



    Dobby.
  • Reply 7 of 24
    yevgenyyevgeny Posts: 1,148member
    Quote:

    Originally posted by the cool gut

    When are we going to finally dump these old ass spinning drives and finally move to solid state storage?



    No time soon. Quit complaining. The thing is that hard drives are going through miraculous growth and there is no reason to switch over to solid state storage unless you are super paranoid about your disks going down in flames. Solid state looses out in terms of cost and storage ability. It just plain doesn't win on economics, so it will be an eternal niche player unless there is some termendous technological breakthrough.
  • Reply 8 of 24
    Quote:

    Originally posted by Yevgeny

    No time soon. Quit complaining. The thing is that hard drives are going through miraculous growth and there is no reason to switch over to solid state storage unless you are super paranoid about your disks going down in flames. Solid state looses out in terms of cost and storage ability. It just plain doesn't win on economics, so it will be an eternal niche player unless there is some termendous technological breakthrough.



    I don't see the point of hard drives going through "miraculous growth" when there is no practical way to back up this amount of data.
  • Reply 9 of 24
    Quote:

    Originally posted by the cool gut

    I don't see the point of hard drives going through "miraculous growth" when there is no practical way to back up this amount of data.



    Absolute tosh, I'm afraid.



    The beauty of SAN and NAS gateway type systems is that you don't have to use tape, because fault tolerance and business continuity are handled by the SAN's inherent ability to deliver a remote mirror to a site many kilometers or miles away.



    The thing to do as data densities increase is to stop clinging to FC over ATM as a viable wide-area mechanism for handling that process, and to use iSCSI or SoIP type concepts combined with QoS and other concepts to take advantage of increases in bandwidth within the Telco environment.



    If you absolutely have to use tape, can I suggest that Super AIT (details of which can be found here with a roadmap that goes up to 4TB/cartridge and 240MB/sec (native) and 10.4TB and 624MB/sec compressed by 2010, is more than capable of coping with disk capacities that double every 24 months.



    At the moment, S-AIT1 gives you 500GB/1.3TB per cartridge and 30/78MB/sec throughput. In other word, a single cartridge will store the contents of around 9 146GB disks when they're formatted.



    If you accept the doubling every 24 months rule, we'll see the first 290GB drives around the start of 2005, 580GB by the start of 2007, and 1.16TB by the end of 2009.



    So a 10.4TB S-AIT4 cartridge will take around 9 formatted disks even in 2010, and will be able to backup a 8TB array in around 3 hours 45 minutes using a single drive and a single cartridge.



    The only thing is keeping it fed with data at just under 5 gigabits/second so the tape is not constantly sawing backwards and forwards, but by 2010, I would imagine that 40Gbit Ethernet technology will be a commonplace as an infrastructure technology, and 100Gbit will be the cutting edge, so 5 gigabits is not really a worry as far as I'm concerned.



    Hope that clears up a few things for you.
  • Reply 10 of 24
    Quote:

    If you accept the doubling every 24 months rule, we'll see the first 290GB drives around the start of 2005, 580GB by the start of 2007, and 1.16TB by the end of 2009.



    Well, if your willing to count ATA drives, then Maxtor's got your timeline beat, their MaXLine II is available at 300GB now (model 5A300J0) for about $350 average selling price online (seen as low are $330). So hopefully we'll have those 1TB+ drives by 2007 or earlier...
  • Reply 11 of 24
    Quote:

    Originally posted by biaachmonkie

    Well, if your willing to count ATA drives, then Maxtor's got your timeline beat, their MaXLine II is available at 300GB now (model 5A300J0) for about $350 average selling price online (seen as low are $330). So hopefully we'll have those 1TB+ drives by 2007 or earlier...



    But server-class SCSI always lags a few years behind desktop-class drives, and FC-AL is even further behind that.



    I do accept that the 1TB SATA may well be a reality by 2007, but the backup scenarios for those machines will be massively different.
  • Reply 12 of 24
    Server SCSI drives are often not the same as desktop ones.



    It is much more efficient to add more disk arms than to increase capacity of the drives, it all depends on what type of an application you are running.



    If you have huge disks and many data requests, your disks will be busy and in turn users will be waiting for their data. If you would have more smaller disks, data requests would spread among more disk arms.



    My only question is why for the love of god are these things soooo damn

    expensive?!?!?
  • Reply 13 of 24
    Quote:

    Originally posted by piwozniak

    Server SCSI drives are often not the same as desktop ones.



    It is much more efficient to add more disk arms than to increase capacity of the drives, it all depends on what type of an application you are running.



    If you have huge disks and many data requests, your disks will be busy and in turn users will be waiting for their data. If you would have more smaller disks, data requests would spread among more disk arms.



    My only question is why for the love of god are these things soooo damn

    expensive?!?!?




    When I started out in this business back in 1984/5, I purchased a RAM upgrade for my employer's Data General Nova 4/X for £3500 (or $5600) - which sounds reasonable until you hear that the RAM upgrade was for 256KB.



    Only a couple of years later, DG were selling a "huge" Winchester disk sub-system called Argos for £73,000 - which bought you a tremendous 354MB enclosed in a massive free-standing unit containing 14" disks.



    So you'll forgive me if I find it hard to get excited now that you can get 146GB (500 times as much) for around £2000 (3% of the cost).
  • Reply 14 of 24
    heh, i guess you are right, expensive is relative term..



    Recently we were upgrading our AS/400s and these HDs aren't cheap..



    but then again it's IBM and you get what you pay for.



    :-)
  • Reply 15 of 24
    Quote:

    Originally posted by Mark- Card Carrying FanaticRealist

    Absolute tosh, I'm afraid.



    The beauty of SAN and NAS gateway type systems is that you don't have to use tape, because fault tolerance and business continuity are handled by the SAN's inherent ability to deliver a remote mirror to a site many kilometers or miles away.



    The thing to do as data densities increase is to stop clinging to FC over ATM as a viable wide-area mechanism for handling that process, and to use iSCSI or SoIP type concepts combined with QoS and other concepts to take advantage of increases in bandwidth within the Telco environment.



    If you absolutely have to use tape, can I suggest that Super AIT (details of which can be found here with a roadmap that goes up to 4TB/cartridge and 240MB/sec (native) and 10.4TB and 624MB/sec compressed by 2010, is more than capable of coping with disk capacities that double every 24 months.



    At the moment, S-AIT1 gives you 500GB/1.3TB per cartridge and 30/78MB/sec throughput. In other word, a single cartridge will store the contents of around 9 146GB disks when they're formatted.



    If you accept the doubling every 24 months rule, we'll see the first 290GB drives around the start of 2005, 580GB by the start of 2007, and 1.16TB by the end of 2009.



    So a 10.4TB S-AIT4 cartridge will take around 9 formatted disks even in 2010, and will be able to backup a 8TB array in around 3 hours 45 minutes using a single drive and a single cartridge.



    The only thing is keeping it fed with data at just under 5 gigabits/second so the tape is not constantly sawing backwards and forwards, but by 2010, I would imagine that 40Gbit Ethernet technology will be a commonplace as an infrastructure technology, and 100Gbit will be the cutting edge, so 5 gigabits is not really a worry as far as I'm concerned.



    Hope that clears up a few things for you.






    I was talking about the consumer market.
  • Reply 16 of 24
    Quote:

    Originally posted by the cool gut

    I was talking about the consumer market.



    I want to meet the "consumer" with 1TB of data - given the way AAC, MP3 or even Pixlet or WMA/V compress data, 1TB implies the ownership (lease if you use buy.com) of a lot of media assets.



    S-AIT1 will ship at around $10,000 for a stand alone unit next spring, so by 2010, it might carry a price tag of around $2500 which doesn't strike me as excessive if you're talking about having assets that might cost you $25000 or more to replace.



    Sony are also claiming that standard AIT will be at AIT-6 by 2007 with 800GB native (presumably just over 2TB compressed) writing at 96MB/second, which strikes me as likely to be a more viable consumer unit at a more affordable price ($1750.00 would be my guess) with the potential to secure a 1TB disk in under 90 minutes.
  • Reply 17 of 24
    Quote:

    Originally posted by Mark- Card Carrying FanaticRealist

    I want to meet the "consumer" with 1TB of data - given the way AAC, MP3 or even Pixlet or WMA/V compress data, 1TB implies the ownership (lease if you use buy.com) of a lot of media assets.



    S-AIT1 will ship at around $10,000 for a stand alone unit next spring, so by 2010, it would carry a price tag of around $2500 which doesn't strike me as excessive if you're talking about having assets that might cost you $25000 or more to replace.




    LOL -so how is "todays" consumer [ work with me - we're in the present buddy] with a couple 180 gig hardrives with home movies, photos and other media are supposed to economically back up their data.



    I must add that your incredible ability of quoting sales brochures and spec sheets is most impressive.
  • Reply 18 of 24
    Quote:

    Originally posted by the cool gut

    LOL -so how is "todays" consumer [ work with me - we're in the present buddy] with a couple 180 gig hardrives with home movies, photos and other media are supposed to economically back up their data.



    I must add that your incredible ability of quoting sales brochures and spec sheets is most impressive.




    It's just research which I have to use to support my clients, seeing as when I spec systems and technology I like to know where it's all going before I spend $10,000 and more of my customer's cash, and it's why I never spec DDS for customers.



    Today's consumer with 180gig drives and fairly standard requirements has to recognise that backups are not a luxury but a prerequisite, akin to buying house insurance - I think you missed my point about the consumer with 1TB of disk space; if iTMS carried on growing in terms of catalog and the model was adapted for movies, games, etc, it would be perfectly feasible for a real anorak to spend $6000 a year for four years, which makes spending $1500 on a backup unit look like a good investment to me.



    There was a horror story on one of the sites this week about someone who bought a whole load of stuff from iTMS whilst a US resident, moved to Canada, had a disk crash and now can't recreate his library - bet he wishes he'd joined .Mac. or did a backup.



    Personally, I wish optical would develop more quickly for today's consumer, because 4.7GB on a DVD doesn't really work in your scenario. Really we need Blu-ray or DVD-HD to come along and make backups affordable and managable for all, either that or we need xDSL or some other always-on link to go up by an order of magnitude to 25Mbps, so that your backup is actually delivered on someone else's systems - funnily enough, this is the theme of a white paper I'm writing at the moment.
  • Reply 19 of 24
    drboardrboar Posts: 477member
    What about FireWire drives as backups? The sizes aviable is way above 4.7 GB as a matter of fact they tend to mach the sizes of internal drives...
  • Reply 20 of 24
    Actually SAIT1 will be available in September in the UK... Consumers who want to make backups of their disks should either invest in an older tape system (ait1,DLT) or buy an external disk and just backup what's needed..

    If you can live with the hassle of re-installing the system and apps if your drive fails. Then backing up data onto multiple DVD's or CD's is not such a big problem..
Sign In or Register to comment.