Become a fan of Slashdot on Facebook

 



Forgot your password?
typodupeerror
×
The Internet

Internet Speed Record Broken (Again) 311

captain igor writes "CNN is reporting that researchers at Caltech and CERN successfully send 1.1 Terabytes of data at a rate of 5.44 Gbps. This is around 20,000 times faster than your typical home broadband connection and almost doubles the previous record. "
This discussion has been archived. No new comments can be posted.

Internet Speed Record Broken (Again)

Comments Filter:
  • I wonder... (Score:4, Funny)

    by inertia187 ( 156602 ) * on Wednesday October 15, 2003 @02:38PM (#7222716) Homepage Journal
    I wonder what they transmitted. Judging by the language in the CNN article, whatever it was, I hope the RIAA or MPAA didn't mind.
    • Re:I wonder... (Score:2, Interesting)

      by jea6 ( 117959 )
      I believe it was "Libraries of Congress".
      • Re:I wonder... (Score:3, Interesting)

        by greenhide ( 597777 )
        modded "Interesting"....*sigh*....

        Mods, think. Is there, in fact, a stack of DVDs you can purchase labelled "Library of Congress, part 1 of 5" etc.?

        No. Whenever lay tech writers talk about data, they describe it in terms of Libraries of Congress, as in, "This new storage format is equivalent to 10 Libraries of Congress" -- which I've always felt is a pretty bullshit quantizer, as the library obviously has things like photographs, movies, and albums that would take a lot of honking space, so much so that n

        • Ah! I get it now!

          You said Libraries of Congress.

          Initially I was confused, thinking it was just "Congress".

          As you know, political bodies emit information in a continuous unending unquantifiable stream, as in

          congress.gov $ nc joevoter.com 80 < /dev/random
        • Re:I wonder... (Score:3, Informative)

          by danila ( 69889 )
          Whenever lay tech writers talk about data, they describe it in terms of Libraries of Congress, which I've always felt is a pretty bullshit quantizer, as the library obviously has things like photographs, movies, and albums that would take a lot of honking space, so much so that no storage medium exists that could conviently and economically store even 1 Library of Congress.

          Sorry to interrupt your crusade against ignorance, but I though you'd find interesting that as early as in 1959 among all people Richa
      • Comment removed based on user account deletion
    • They transmitted only zeros.
  • Yes? (Score:3, Insightful)

    by NetNinja ( 469346 ) on Wednesday October 15, 2003 @02:40PM (#7222737)
    And it will continue to be broken.

    On to other news
    • Agreed. It really depends on how you define the Internet. (As weird as that sounds.) I can walk out tomorrow and (with adequate money, which I unfortunately dont' have...) buy some fiber and run 10 Gbps Ethernet over it. The technology exists today.

      It really depends on where you draw the distinction between LAN and the Internet, but the trend seems to be running everything over Ethernet nowadays. I'm waiting for someone to bring a new 10GigE backbone online and run something like this.

      5 Gbps really isn't
  • by wmaker ( 701707 )
    and is also equivalent to transferring a 60-minute compact disc within one second -- an operation that takes around eight minutes on standard broadband.

    What broadband is this? my cable modem can't download 600 megs of data in 15 minutes.
    • When I was living in the dorms at osu I'd get 1024 kbps sustained during off peak hours. Just depends on the connection I guess.
    • Yeah, I noticed that too. That would be over 10Mbps, right? I've never heard of any cable modems that fast.

      Whenever I've tried to download Linux ISO's, even over a full T-1, I've normally had to wait more than an hour each.

      I guess it boils down to "what is broadband?" I guess a T1 must not fit their definition anymore.

      --
      Slash
    • Their math is wrong all over the place. 1) 20,000 times faster that standard broadband 2) "compact disc within one second -- an operation that takes around eight minutes on standard broadband". 8 minutes = 480 seconds, i.e. 480 times faster. This is a difference of about a factor of 40 from the 20,000 number. Also: 90-minute DVD download in 15 minutes. Assuming 3GB (shooting lower than a full DVD here). That would be 25Mbit/second. That cannot refer to "standard broadband" but neither to the previous speed
    • That comes down to about 685kB/s. You can get that on an uncapped ADSL connection if you're close to the exchange. If I get a good source, I can manage 800kB/s *huge grin*
  • by Medcoop ( 687233 ) on Wednesday October 15, 2003 @02:41PM (#7222754)
    Until I can download a pizza in 30 minutes or less, I will not be satisfied!
    • That might actually be interesting to think about when we get printing devices that make objects rather than paper printouts. (I'm talking about depositing materials, not cutting away as in a lathe.)

      Anyone know of any good discussions on atomic-level object imaging?
  • News Flash (Score:3, Insightful)

    by MachineShedFred ( 621896 ) on Wednesday October 15, 2003 @02:42PM (#7222767) Journal
    Technology improves over time.

    Why don't I just die from suprise? At least THAT would be news.
  • by stanmann ( 602645 ) on Wednesday October 15, 2003 @02:42PM (#7222775) Journal
    With this, I'll be able to fill up the IBM Storage Tank I ordered... I want to download the internet...
  • This is around 20,000 times faster than your typical home broadband connection and almost doubles the previous record.
    20,000 you've obviously never used an NTL connection. I can only dream of low pings and quick streams!
    • LOL I came into this thread with every intention of posting a derogatory reference to NTL but you beat me to it!

      Plus DNS never works either.

      Any UK /.'ers recommend a good DSL provider?

      • I've had very few problems with Pipex. Doesn't seem slow either, but maybe thats cos not many people use broadband around these parts.
        • Couple of friends have recommended Pipex.

          I guess the problem is these DSL providers have the same issues of popularity / demand / scalabilty as the ol' days of unlimited dial-up.

          As soon as any particular service got a good reputation; thousands signed up; thus rendering the service overloaded and useless.

          I guess from the A&A website that they are trying to avoid mom&pop type customers on purpose; perhaps to save on support costs. They are a bit pricier than freeserve/openwoe etc. etc.
  • EVERYBODY (Score:3, Funny)

    by Doomrat ( 615771 ) on Wednesday October 15, 2003 @02:43PM (#7222791) Homepage
    Shut up about porn.
    • by Pharmboy ( 216950 ) on Wednesday October 15, 2003 @04:13PM (#7223679) Journal
      Shut up about porn.

      Interesting comment. Allow me to demostrate why we all owe a debt to porn, and the people who enjoy it.

      Back when VRC's first came out, they typically ranged from $1200 to $1800 each. The reason you can buy one for $49 now is because of porn and those who enjoy it.

      When video cameras first came out, they were the size of suitcases and cost in excess of $3000 each for mediocre quality. Now you can buy an exceptional quality unit for well under $1000, and get a decent model for closer to $500. This is because of porn and those who enjoy it.

      When computers first came out, growth was slow and mainly due to business' wanting to automate. Once computers became powerful enough to be useful for full graphics (386dx) they were still terribly expensive ($2000 for a stripped box) but soon came down because of porn and those who enjoy it.

      When the Internet first became available to the general public, in mass, I personally paid $80 a month for 80 hours. (ironically, not for porn, but for a BBS). Many before me paid much more. But now you can get dialup for $10 to $20 a month, because of porn and those who enjoy it.

      Like any new technology, the price can only come down once two conditions are met:

      1. Demand is high enough at inflated prices to pay for the research and development involved in bring out a new product. This allows a company to recover a portion of their original investment.

      2. Demand has to be reliable enough for companies to invest in excess manufacturing capacity. This lowers prices because it introduces economy of scale. It introduces competition because any profitable venture will attract capitalists who want to make a better mouse trap, cheaper. Eventually, it turns the new product into a commodity, where margins are razor thin and you can get the same basic product from a number of providers.

      In each case, it was porn and those who enjoy it that invested the money on the consumer side for these products. No one would have paid $3000 for a computer to email someone on a $80 a month 2400 baud connection. (think fidonet or google it) No grandmother would have paid $3500 for a video camera to take shots of her grandchildren, to play on her $1800 VCR.

      Microwave ovens became available in the 50s, but they did not become popular until the late 70s. Why? Because they have no porn value, so it took 20 years to get the economy of scale and demand strong enough for the price to come down. Had there been a potential porn use for the microwave, we all would have had them for $100 before we landed on the moon.

      We all owe a great debt to those brave pioneers, who worked tirelessly typing with one hand, pants to ankles, in the darkness of night. Because of their relentless pursuit of a better way to masturbate, we are all able to enjoy consumer goods at incredibly cheap prices. Even the third world countries are able to benefit with wireless phones where there are no wires, all because some guy sitting in his parents basement was patient enough to wait for a 256 color GIF image to load to screen over a 14.4kbit connection.

      So, the next time you see a pervert, go up and shake his hand, and tell him "Thank you for your contribution to society".

      Just be sure to wash your hand afterward.

      • Microwave ovens became available in the 50s, but they did not become popular until the late 70s. Why? Because they have no porn value, so it took 20 years to get the economy of scale and demand strong enough for the price to come down. Had there been a potential porn use for the microwave, we all would have had them for $100 before we landed on the moon.

        Someone give this man a Nobel Price in Economics!

        +1 Funny and Insightful.

  • My DS3 looks .... Slow.

    The real application for this kind of speed is "Click and Watch" (TM)(C) Movies. What point is there to downloading if it is nearly instantaneous delivery?
  • by i.r.id10t ( 595143 ) on Wednesday October 15, 2003 @02:45PM (#7222813)
    And figure out if this still beats a station wagon (or SUV or whatever) loaded with DVDs, CDs, backup tapes, etc.
    • Assuming a DVD weighs approximately 1 lb (Case included) and max weight for a legal semi is 80,000 lbs; max load is around 45,000 lbs of cargo, meaning

      45,000 DVD's x 4.7 GB = total data 211,500gb

      Traveling at 70 miles/hour for say 1000 miles = ~14.3 hours traveling

      14790 gb/hour
      246gb/minute
      4.1gb/second/70 miles hour/1000 miles/45,000 DVD's

      • by Raul654 ( 453029 )
        You forget the overhead required to generate the payload - packing and unpacking all those DVDs would take a LONG time.
        • How long did it take to get the systems ready and initiate the transfer?
          My calculations also assume perfect travel conditions
    • A 2002 Ford Expedition features 106 cubic feet of storage space with the 3rd seat removed and the 2nd row seats folded. A DVD in a standard jewel case is approximately 5 5/8" square and 3/8" thick. So we can safely assume that 32 DVD's (in jewel cases) could be stacked in a one foot pile and that four such piles could be placed in a square foot area. This gives us 128 DVDs per cubic foot, and 13,568 DVDs per Ford Expedition.

      The capacity of a DVD-18 is 17 GB [techtv.com]. So, our Ford Expedition has the capability of h

  • the last mile (Score:2, Interesting)

    by dilvie ( 713915 )
    The real question is, when are we going to have better speeds for home users? Even "broadband" connections are slow. Is there any progress being made in this arena right now? Perhaps faster data transfers over cable lines?
    • Broadband connections ARE getting faster. I stopped my Adelphia service for a few months while I was out of town, when I started it back up, they said they had just finished a network upgrade, and my avg. DLS reports speed test went from 1.5 Mbit/s to 2.2 Mbit/s. It's not an amazing gain, but figuring that there are more people in my area running from the same connection now, and speed increased, I'm pretty happy.
  • by 192939495969798999 ( 58312 ) <info.devinmoore@com> on Wednesday October 15, 2003 @02:46PM (#7222824) Homepage Journal
    I transmitted a dictionary across the room in .05 seconds, when I threw it. I think it's important to note the type of connection that they are using, protocol, etc... hardware? software? C'mon, guys! Post something in the article that lets us know some detail, so that I know it wasn't just a dictionary being thrown across the room or something dumb like that.
  • us perverts await our vr immersive experience
  • Yes, but (Score:3, Interesting)

    by dustmote ( 572761 ) <fleck55&hotmail,com> on Wednesday October 15, 2003 @02:47PM (#7222840) Homepage Journal
    I still want to get off of dialup at my apartment. And even when I had broadband, there were still sites that wouldn't load very quickly. The servers are going to need some upgrading as well, I think, before bandwidth becomes the only bottleneck. Still, that's really cool. I hope to see something approaching instant response within my lifetime. Besides my old DOS computer, way back when. :)
    • Bandwidth hasn't been the only bottleneck for quite some time. Having high throughput between two points on a connection doesn't mean you have hige throughput across the entire connection. If somebody's server is at the end of two tin cans and a bit of wet string, it doesn't matter how fast your local connection is...
  • by 4of12 ( 97621 )

    So this other related recent accomplishment [sciencedaily.com] must just be chopped liver at only 6.8 Gbps, then?

  • Argh (Score:4, Funny)

    by Kelz ( 611260 ) on Wednesday October 15, 2003 @02:55PM (#7222940)
    Ever think about SHARING?!

    While these guys are transferring at 5+Gbps, I'm stuck at home with my 28.8k dialup (no cable/dsl here folks).

    Just like the government studies that cost millions of dollars to figure out why mice will eat cardboard... I can put that stuff to USE other than breaking some damn record!
  • It would be a lot nicer if they had included some technical details on this. 10Gbps links are available, though not common. Indeed, they are presently improving the quality and price of 40Gbps equipment.

    But, how does one drive the data at 5, 10 or 40 Gbps. These speeds are not a big deal for network switching gear but it is a big deal for a PC. The fastest PCI bus that I have seen maxes out a under 5Gbps and there aren't any disk drives that can offer that sort of throughput. Then one has to wonder how the
  • ... among sysadmins that translates roughly as "never underestimate the bandwidth of a crate of tapes in the back of a station wagon".

    These CERN upstarts don't impress me much, I had and was using more capacity a decade ago!
  • Does anyone know how they did this? I mean, is the future of broadband-at-home going to be one fiber-optic (or faster?) cable or are we all going to have x ethernet cables hooked up to our computer?

  • The original press release is here [uic.edu].

    Is TCP's performance really that poor? Some UDT presentations quote 2.4 MBytes per second. Over a low-latency WAN (few dozen milliseconds), performance is actually quite good, and sometimes, it used to be faster to fetch a file from a site a few hundred kilometers way than from the local FTP server (although the latter was connected to the LAN using a 100 MBit link).
  • Yes, but... (Score:3, Insightful)

    by jd ( 1658 ) <imipak&yahoo,com> on Wednesday October 15, 2003 @03:01PM (#7223014) Homepage Journal
    ...when will my ISP provide these kinds of speeds?


    It's time to face facts. "Broadband" isn't, and won't be, until we're at least at the 1 Gbit/s rate to the home. In fact, with gigabit cards starting to become affordable, and with home networks on the rise, a gigabit link to the house may not be fast enough in only a few years.


    Running a modern PC over so-called broadband networks is like towing a Ferrari F1 car using a couple of Shire horses. Sure, it "works"...


    For the money so far spent on rebuilding Iraq, the US Government could have built a network of 2 terabit lines between every pair of States in the US, installed the clusters of routers needed to handle the load, and provided lines to every carrier of Internet and phone traffic in the country. They'd probably still have cash left over.


    This isn't to say we shouldn't rebuild Iraq. This is very much to say that if organizations and Governments can throw that kind of cash around as though it were spare change, then I'd really like to see some serious infrastructure upgrades in a certain country whose economy and security both need those upgrades to take placed.

    • It's time to face facts. "Broadband" isn't, and won't be, until we're at least at the 1 Gbit/s rate to the home. In fact, with gigabit cards starting to become affordable, and with home networks on the rise, a gigabit link to the house may not be fast enough in only a few years.

      The problem being that a 1 Gbit card easily outpaces your hard drive, your PCI bus and is able to stream pretty much any media several times real-time without any problem. There is no need for gigabit to the home right now. A 100 M
  • by tessaiga ( 697968 ) on Wednesday October 15, 2003 @03:04PM (#7223044)
    Steven Low [caltech.edu] of CalTech's Netlab [caltech.edu] gave a talk [mit.edu] at MIT yesterday regarding the modified TCP protocol they used to achieve this transfer. Those who are curious about the details can check out the Fast TCP [caltech.edu] homepage.

    Basically they showed that conventional TCP is not very good at scaling to large flows like the ones in the article. He described a typical broadband Internet connection as being able to utilize only about 27 percent of the available bandwidth, while their modified FAST TCP connection reached 95 percent efficiency. He had some nice test results showing how the protocols reacted to having to share bandwidth with other flows, and pointed out how when other flows finished and more bandwidth opened up, conventional TCP was very slow to take advantage of the increased bandwidth.

    There's an older Economist [economist.com] article describing the protocol in more detail for those who are interested.

  • 5.44 Gbps is nothing. I recently took a box with about 300 CDs in it to my neighbors place. It took about 10 seconds to walk there, giving me a throughput of (650MB * 300 * 8) / 10s = 156 Gbps.

  • For everybode with their RIAA jokes, i guess this kind of work is used to pave the way to use the LHC.
    With Petabytes of Data each year, a normal internetconnection simply doesnt cut it.

    Want to give the data of a single experiment to some guys on the other side of the atlantic? Just send 100GB...

    A multi-petabyte storagenet like the proposed storagetank does only make sense if the infrastucture allows to actually transfer the data with such speed.
  • A 747 has a cargo of 232,000lbs
    Assuming DVD is 4.7GB, ~ 1lb
    232,000 DVD's is 1,090,400GB
    Speed of 570 miles/hour
    Distance is 5700 miles (Berlin to LA, so close)
    10 hours of flight time

    109,040GB/hour
    1817GB/minute
    30GB/second

    the 747 still wins (not counting loading and fueling times)
  • I am no hard drive expert, but even on my 15K RPM scsi drives, I am not sure I could write 1 terabyte in 30 minutes. What are they doing with the data on the other end?
  • by 4/3PI*R^3 ( 102276 ) on Wednesday October 15, 2003 @03:11PM (#7223116)
    Cop:Sir, do you know how fast you were going?
    User:Ummm, I'm not sure my speedometer has started messing up. It felt like I was going about 256 Kbps.
    Cop:No sir, I clocked you at 5.4 Gbps. Thats 20,000 times the speed limit. You blew past me like I was in reverse.
    User:Gee, officer it must be this new European packet switching system I've added to my cable modem.
    Cop:Tell it to the Judge. MAC and IP adddress please...

  • holy cow -- that many Gbps is faster than the time it takes your brain to determine whether you're looking at shot of tiffany vionette or a lactating naughty grandma!

  • Now they can actually keep up with downloading required Windows patches.
  • There are cable-laying ships laying cables in all the seas out there to make distance carriers rich. Each cable is a big bundle of optic fibres carrying ip over ATM traffic between juniper routers. Each cable also carries somewhere between an OC-48 to OC192 traffic. This really makes me wonder:

    (1) Why is 5gbps a record? Why is it not possible to connect OC48-supporting ATM or FDDI PCI-64 cards on both ends to servers and then mirror some important servers carrying all the free OSes (like ibiblio.org) to co
  • The problem is that these transfers were not done using general purpose protocols. They are all highly tweaked protocols, even if TCP/IP, that are customized to the hardware and pipes the data will transfer over. This will never be feasible for a home user. It's all just for show, it's not practical.
  • Oct 10th: 6.8Gb/s (Score:2, Informative)

    by Tangent55 ( 189500 )
    University of Illinois at Chicago was able to achieve 6.8Gb/s a few days ago using the UDT protocol .... http://www.ncdm.uic.edu/pressrelease.html [uic.edu]
  • Firehose [heroinewarrior.com]gives you that power. FIREHOSE gives you a basic data transfer over multiple network devices supporting TCP/IP layers. Stripe multiple 100Mbit, Gigabit, 10 Gigabit, or firewire to give one humungous pipe for firehosing your gigabytes and gigabytes of data
  • Bit Torrent (Score:3, Funny)

    by DeadBugs ( 546475 ) on Wednesday October 15, 2003 @03:50PM (#7223501) Homepage
    • "1.1 Terabytes of data at a rate of 5.44 Gbps."
    Does anyone have a bit torrent link to that file?
  • Step aside Caltech and CERN, your record of 1.1 Terabytes of data at a rate of 5.44 Gbps has already been broken. From this article [sciencedaily.com]:

    UIC's National Center for Data Mining (NCDM) and Laboratory for Advanced Computing flashed a set of astronomy data from Chicago to Amsterdam at 6.8 gigabits per second

    and

    The test used Amsterdam's SURFnet and Chicago's Abilene networks. During a 30 minute test, the researchers transmitted approximately 1.4 terabytes of data

  • CA*Net 3 supposedly operates at up to 40Gbps [canarie.ca], with CA*Net 4 under development which should be four to eight times faster. It's not clear, however, whether this is an aggregate data rate or if it can be sustained on a single connection.

    Regardless, we'll eventually have Tbps data rates and all this will be a moot point. I only hope that a spammer doesn't manage to get one of those connections. How many viagra and penis spams per second is that? :)
  • CERN, whose laboratories straddle the Franco-Swiss border near Geneva, said it had sent 1.1 Terabytes of data at 5.44 gigabits a second (Gbps) to a lab at the California Institute of Technology, or Caltech, on October 1.

    Finally somebody spelled Caltech [caltech.edu] correctly. It's not Cal Tech or Cal-tech, and it damn sure isn't CalPoly [calpoly.edu] or PCC [pasadena.edu].

"And do you think (fop that I am) that I could be the Scarlet Pumpernickel?" -- Looney Tunes, The Scarlet Pumpernickel (1950, Chuck Jones)

Working...