Forgot your password?
typodupeerror
Communications Technology

Siemens Reaches 107 Gbps Data Transfer Record 161

Posted by samzenpus
from the greased-lightning dept.
prostoalex writes "Reuters is reporting on Siemens engineers reaching 107 Gbps data transmission record over a fiberoptic cable, and expects the technology to be on the market within a few years: "The test, 2.5 times faster than a previous maximum transmission performance per channel, was done in cooperation with Germany's Micram Microelectronic, the Fraunhofer Institute for Telecommunications and Eindhoven Technical University of the Netherlands.""
This discussion has been archived. No new comments can be posted.

Siemens Reaches 107 Gbps Data Transfer Record

Comments Filter:
  • Hooray! (Score:4, Funny)

    by PixieDust (971386) on Wednesday December 20, 2006 @11:48PM (#17321622)
    And everywhere, lonely geeks rejoice at the decreased download time for the favorite pr0n!
    • The problem is... (Score:4, Informative)

      by TrisexualPuppy (976893) on Wednesday December 20, 2006 @11:51PM (#17321654)
      And everywhere, lonely geeks rejoice at the decreased download time for the favorite pr0n!
      This will not matter much, at least on the individual's machine. Most hard disk drives transfer on the order of 25MB/s. This fiber transfer is applicable only for supercomputing links and Internet backbones. Good luck finding a 107000000kbps stream ;)
      • Re:The problem is... (Score:5, Informative)

        by Barny (103770) <bakadamage-slashdot@yahoo.com> on Thursday December 21, 2006 @12:00AM (#17321710) Homepage Journal
        Yes it will actually, this is a bigger pipe, and since the internet is a series of pipes..

        But really, if an Aussie ISP (internode for instance) has just upgraded from 3Gb/s to around 6Gb/s, how much would it benefit them if they could just sell off most of the fibre they are using currently and just run one at 107Gb/s?

        As for 25MB/s, a newer HDD will easily reach around 40-50MB/s, added with the popularity of NAS and small raid systems most good PCs can suck almost 70MB/s (560Mb/s).

        Of course, with Australian broadband being lucky to get (until just recently) above 1.5Mb/s this is rather moot.
        • by Mike89 (1006497)
          Hello, fellow Whirlpoolian ;)
        • by cciRRus (889392)
          Yes it will actually, this is a bigger pipe, and since the internet is a series of pipes..
          I think you meant tubes [wikipedia.org] instead.
        • by s31523 (926314)
          Or you could use the HyperOS HyperDrive [hyperossystems.co.uk], which is a DDR RAM based drive which claims "Seek time is 50-60 microseconds", that is right microseconds.

          Granted, this is RAM based technology with the obvious pitfalls (this one has a external power source so as to not lose your data), but with upcoming flash based drives and other improvements, I wouldn't be surprised if these drives become more mainstream, especially with network transfer rates of 100Gb/s.
      • This will not matter much, at least on the individual's machine. Most hard disk drives transfer on the order of 25MB/s. This fiber transfer is applicable only for supercomputing links and Internet backbones.

        And my ISP is a gateway to those lines. I'd rather be pushing my data through someone else's big tubes than my own... its a lot cheaper that way.

        Good luck finding a 107000000kbps stream ;)

        Well, I might be able to find a 25MB/s stream ;) and that's good enough for me.

      • Re:The problem is... (Score:4, Informative)

        by NerveGas (168686) on Thursday December 21, 2006 @12:10AM (#17321764)

        Most hard disk drives transfer on the order of 25MB/s

        Maybe you should upgrade that machine you bought four years ago. :-)

        A lot of drives today can write at twice that speed, and read even faster. I've got an external firewire 800 drive (a single drive, not one of the RAIDs-in-a-box setups) that can write at a little over 60 MB/s. Your point is, of course, still valid... few users are even able to make use of a gigabit - or sometimes even half of that.

      • Way to kill the guy's joke. Just let him enjoy his +2 funny. :-P
      • Even if the report is true it does not descibe any major feat - no reason to get excited then.

        But if what one can hear in German media about Siemens corruption is at least partially true, then one may start having serious doubts whether the results are real or the report is bogus 'cause it was bought by money saved in e.g. BenQ Mobile [wikipedia.org] disaster (more details only in german version of the article I am afraid). For those that missed the story: Simens sold its mobiles making division (together with people) to

        • It is a huge feat. Its 107gb per second per channel. That means one can layer on multiple channels through a single fiber, for outrageous throughput. When many computers have to talk to many computers over a single run of fiber (say, like on the internet across intercontinental links) this kind of throughput is can be kind of important.
      • by drsquare (530038)
        HD limitation is only an issue if you're saving a file. It's not important for watching videos. But at the end of the day your speed will be limited by all the links from your computer to the site. I have an 8MB connection but still can't download from youtube faster than it plays. I can on Google video, but then there's not much to watch on there.
        • Yeah, but this is getting up there with your frontside bus bandwidth.
          • Re: (Score:2, Funny)

            by bdonalds (989355)
            Yeah, but this is getting up there with your frontside bus bandwidth.

            Aren't you listening? This technology allow you to download more porn, which increases the width of your frontside bus!
      • by Kjella (173770)
        Good luck finding a 107000000kbps stream ;)

        Well, if you're working with raw HDTV then a HD-SDI link is a 1485000kbps stream, so about 72 of those at the same time should do it. Ok, so it's not everyday use but for individual's machine: I doubt you could fit all the cameras of the Olympics to a master console through one of these cables. In any case, it doesn't take much of a town before you can saturate this cable.
      • by pclminion (145572)

        Interesting you mention it. My wife works at a very large maker of computer parts (which will remain unnamed). They regularly have to send several terabytes of technical data between a site in Oregon and one in California. They have a FAT PIPE to do this with, but they discovered that the servers on either end of the pipe aren't fast enough (sending OR receiving) to saturate the pipe.

        So my wife's task was to construct a program which reads the data in parallel from their massively parallel filesystems, co

      • by Anthony (4077) *

        Unless you have access to SAS (Serially Attached SCSI) drives :)

        sudo /sbin/hdparm -t /dev/sdb

        /dev/sdb:
        Timing buffered disk reads: 268 MB in 3.00 seconds = 89.20 MB/sec

        sudo /sbin/hdparm -T /dev/sdb

        /dev/sdb:
        Timing cached reads: 12632 MB in 2.00 seconds = 6321.11 MB/sec

        Or even better, a HPC cluster file system with Fiber Channel and/or SAS drives

    • Re: (Score:3, Funny)

      by poopdeville (841677)
      107 Gb/s sounds like a lot. How much is that in Metallica discographies/s?
      • Re: (Score:2, Informative)

        by Jello B. (950817)
        About 110 Metallica discographies per second, according to a torrent I found, which lists it at 973 megabytes. That should be a new file transfer measurement. Md/s.
        • In other words, it's competitive with CDs in the mail. Especially if you have to put them in the drive and read them in the other end.

          I'm impressed.
          • by oc255 (218044)
            Huh? I don't think anyone could snail-mail CDs fast enough without using very large bulk shipments. You figure:
            - 2 day air shipping by specialized carrier
            - an unlimited amount of CD drives all operating at the same time would still take some time to maintain, load, setup, etc even with unlimited staff

            So let's call it three days. Using GP's 110 Md/s number, I'd have 259,200 Metallica albums transferred at this speed in a digital format, of course I suppose the filenames and album names would all be duplic
        • Re: (Score:2, Funny)

          by 42Penguins (861511)
          Not to be confused with MD/s, Megadeths per second, a much smaller unit.
    • So maybe "Siemens" wasn't quite the word they were looking for?
  • See also (Score:4, Informative)

    by Sampizcat (669770) on Wednesday December 20, 2006 @11:53PM (#17321666)
  • I wonder if we will get higher speeds on copper or maybe just cheaper fiber interface cards. Fiber optic networking technology has always been fast, but I guess due to production quantities, it never seems to be as cheap to implement even in a Data center environment. I wonder if we will ever get to see fiber optic network interfaces that are close in price to the copper ones.

    We run multiple cat6 cables as trunk links between our switches just because there are more ports to do so and it is cheaper to do th
    • by timmarhy (659436)
      huh? fiber isn't expensive when your talking data centers...
    • errr... I almost bet your switches are 4m appart. 10m? I've even say 20m. But not kms appart, which is what these systems are for.
  • Excellent, but... (Score:2, Insightful)

    by JoshJ (1009085)
    If this technology is proprietary (which it likely will be), the lock-in could be rather vicious, especially for the businesses that would likely be the first ones impacted, and when it does get around to the average citizen, they could give horrible service, drop people, restrict their bandwidth, etc- and they'd be able to get away with it because of the monopoly they'd get over the high speed.
    • Re: (Score:2, Funny)

      by Anonymous Coward
      If sun would shine and it wasn't cloudy, this could be a warm day. I bet you didn't know that. Now, mod me insightful.
    • by modecx (130548)
      Huh?

      So you're suggesting that Siemens AG will able to say "okay bend over" to everyone and his brother, because they're going to suddenly be solely in control of this technology, no less the rest of the global telecom industry, and everyone will put up with their asinine ways because bandwidth will fall from the skies like mana from heaven?

      Suuuuure....

      Here's how Siemens works: They make stuff and they sell stuff, and as much as they like to buy other companies, it's just not in their usual habit to buy all
  • by sycodon (149926) on Thursday December 21, 2006 @12:05AM (#17321742)
    ...fifteen looong seconds to list the contents of a folder.
  • How much effort and cost is involved with upgrading the current backbones to this standard? Can existing fibre be used? Especially all the "dark fibre" that was laid during the .com boom and AFAIK is still just sitting there unused to this day! If existing fibre can be kept only having to upgrade optical nodes could provide a relatively cheap upgrade to network bandwidth in the US at least?
  • by Anonymous Coward
    "The Fraunhofer Institute for Telecommunications"

    I assume that's related to the institute that gave us the "proprietary" MP3?
  • From the article:

    sent it over a single optical fiber channel in a 100 mile-long (161-kilometre) U.S. network

    After 100 miles, how much does the throughput degrade? The technology might be limited if, after 200, 500, or even 1000 miles, its speed drops significantly. Or does it reach a hub of some sort that re-sends the signal every 100 miles? I should admit now that I'm not very familiar with how large telecom networks are set up.

    • by MrJynxx (902913) on Thursday December 21, 2006 @12:49AM (#17321968)
      Telecom companies dont' usually have fiber that long because of the risk of breaks and really costly repair processes, it's not because of degradation. Also the distance doesn't really matter(remember, how do you think the contients are connected? single link fiber), because if it's a good cable the data should travel at the speed of light. It depends on the recieving ends how fast your can process it.

      Also the infrastructure for telecom is quite large, you'd be surprised how much stuff is running underground.
      • Re: (Score:3, Insightful)

        by 3.14159265 (644043)
        sigh... of course they usually have fiber, that's the only way you've got to carry those kind of bitrates! What do you think Verizon and AT&T are getting? CAT5e?
        sigh... of course the distance matters, the higher the span length the higher the attenuation and dispersion!
        sigh... if they say they can do 107Gb/s that's because they can fire up the laser on one side and get it with an acceptable bit error rate at the other side. These tests are not based on sending something to /dev/null!
      • by umghhh (965931)
        I am part of this infrastructure and I am not running. Not under the ground anyway.
      • Re: (Score:2, Informative)

        by Anonymous Coward
        Okay there seems to be some misinformation that I must correct here.

        Distance is very very important in fibre systems. Distance causes attenuation (which affects the signal-to-noise ratio), polarisation mode dispertion, chromatic dispertion etc. All of these have a detrimental affect to the bit-error-rate at the reciever. All must be compensated for along the way. With long reach systems, intermediary nodes are required to regenerate the signal, amplify it, re-shape it, re-time it etc. In addition, lengths o
    • Re: (Score:2, Interesting)

      by ZX3 Junglist (643835)
      This figure of 100mi is actually quite good,
      a general standard in the industry is in the order of 30mi of fiber before signal regeneration is necessary. This is a main reason why the US is not very well suited to fiber octics transmissions in the way a smaller countries like germany or netherlands are. It's not the cost of running fiber, but the cost of maintaining sites and equipment to provide a long distance (cross-country?) signal.
      • What general standard? Nowaday it's pretty common to have 400..500km without electrical regeneration, and certain systems go all the way up to 2500km. Either way, how else are you going to carry traffic?
    • Re: (Score:3, Informative)

      by 3.14159265 (644043)
      The bitrate ("speed") is always the same, no matter the distance. What changes is the bit error rate, which is proportional to the distance. For this particular test they defined a certain bit error rate as acceptable (don't really know which, 10^-15, 10^-16?) and when they say they did 107Gbps over 100km it means they've got the signal on the other side with a bit error rate low or equal to the defined one. When the bit error rate it just too high, you need to electrically regenerate the signal, which is a
  • by Anonymous Coward on Thursday December 21, 2006 @12:18AM (#17321810)
    Then Steve Ballmer can say something like "I can squirt Siemens"
  • A dump truck for of DVDs does at least 5 Terra-bytes a second.
    • Re: (Score:2, Insightful)

      ... Imagine the Blue-ray version! /// ...Imagine the dual-sided Blue-ray version! /// ...Imagine a bewolf.. no wait that doesn't apply, unless it's in Russia.
    • Sounds logic, until you figure you have to burn those DVDs first :)
  • by nick_davison (217681) on Thursday December 21, 2006 @12:50AM (#17321972)
    Record?

    Given the amount of information DNA encodes... that there's, what, a complete set in every single sperm?... I think my Siemen can squirt more than 107Gbps of data per second down "a series of interconnected pipes" than their Siemens can.

    Of course, that's of minimal practical use as a) Those are burst figures, I'm damned if I can sustain them and b) I read Slashdot which means my odds of finding a compatible interface are pretty minimal.
    • by xebecv (1027918)
      Dude, I bet you cannot shoot as far as they did, you have to do a lot of training for this.
    • by alanwj (242317) on Thursday December 21, 2006 @02:36AM (#17322454)

      Given the amount of information DNA encodes... that there's, what, a complete set in every single sperm?... I think my Siemen can squirt more than 107Gbps of data per second down "a series of interconnected pipes" than their Siemens can.
      The bandwidth of a penis [everything2.com] is estimated at 15,600 tb/s.
      • by D4rk Fx (862399)
        The big problems are that it's only one way, and the data content is just random garbage. I mean, look at most of the posters here on slashdot...
      • by Daverd (641119)
        But the latency is terrible.
    • by kestasjk (933987) *
      There are apparently around 3 billion base pairs in our genome, each base pair has 4 states it can be in, which is the equivalent of 2 bits, so that's about 6 billion bits of information per sperm, times 5 million (iirc) ~= 3 TB.

      On the other hand this is the one form of data transmission for which speed from the start of the transmission to the end is not a priority..
      • by High Hat (618572)
        Well, actually the protocol is redundant by factor 5*10^6 because the Link is rather "unstable". As well as a maximum inbound capacity of the receiving end of 6 billion bits. So we're talking about 6 GBit/E.

        Figure out how many seconds E amounts to yourself. :)

    • by Hillgiant (916436)
      The heck with a station wagon full of backup tapes, you should see the bandwidth of MY PANTS.

      Hrmmmm... that is bad on so many different levels. I am sorry to have subjected y'all to it. (But not sorry enough to not hit [Submit])

  • by ksw2 (520093) <obeyeater.gmail@com> on Thursday December 21, 2006 @01:04AM (#17322028) Homepage
    Siemens said in a statement it had processed data using exclusively electrical means at 107 gigabits per second -- roughly two full DVDs per second [...]

    Damn, I can barely keep up with the 5 DVDs at a time I get from Netflix.

  • by the_REAL_sam (670858) on Thursday December 21, 2006 @01:37AM (#17322170) Journal
    I really don't know why they express download speeds in such an outlandish way. End users do not "gigabits" ...gigglebits, maybe, but not gigabits... for anything, they use kB, MB, & GB.

    107Gb/s = "107 gigabits per second"
    13,696 MB/s = "13,696 megabytes per second"
    13.375 GB/s = "13.375 gigabytes per second"

    Source:
    http://www.matisse.net/bitcalc/?input_amount=107&i nput_units=gigabits&notation=legacy [matisse.net]

    Divide by 8 to get the number that makes sense. The "little b" stands for bits, and there are 8 bits per byte; the "big B" stands for byte.

    1B = 8b.

    The byte is the amount of data you could store on a single coin if you had a code worked out placing it either heads up or heads down. Ones and zero's.

    Source:
    http://www.theonion.com/content/node/29130 [theonion.com] :-)
    • Re: (Score:3, Informative)

      by TopSpin (753) *
      End users do not

      End users routinely use multiples of bits per second. Some examples; modems 1200/2400/9600/56k b/s, SATA 1.5/3.0 Gb/s, USB 480 Mb/s, Firewire 400/800 Mb/s, Ethernet 10/100/1000 Mb/s, 802.11b 11 Mb/s, etc.

      Using bytes introduces too much ambiguity when discussing line capacity. In real communications bytes are often encoded (8B/10B) or are accompanied by (a possibly configurable number of) error correction bits. Higher level protocols add effectively arbitrary amounts of overhead. People w
      • Although what you say is true, it does not change the bottom line: bytes and bits are interchangeable measures of the same thing. If potentially inefficient high level protocols, parity bits and SSL are taking that 15% from bytes per second then they're also taking 15% from bits per second.

        Right? :-)
    • The byte is the amount of data you could store on a single coin if you had a code worked out placing it either heads up or heads down. Ones and zero's.

      almost

      The bit is the amount of data you could store on a single coin if you had a code worked out placing it either heads up or heads down. Ones and zero's. A byte would therefore need 8 coins.
      • Amazing. I can't believe I mistyped it, but you're right. Thanks for pointing it out. I'd hate to misinform people.
    • by dtzWill (936623)
      Good point, and I agree. However:

      The byte is the amount of data you could store on a single coin if you had a code worked out placing it either heads up or heads down. Ones and zero's.
      A bit is the one or the zero, or the 'heads' or 'tails' in your example. And as you correctly stated, a byte is 8 bits.
    • by Detritus (11846)
      Engineers and scientists use bits and symbols. Take a look at any text on the mathematical theory of communications. Bytes are ambiguous (see octet) and are at a higher level of abstraction. While we're at it, k = 10**3, M = 10**6, G = 10**9.
  • I really couldn't care less until I can get greater than 486kbps over my ~17,000 foot link to my CO. You can transmit your butt to Uranus at the speed of light and it isn't going to make me thing you're some kind of super hero. That would still be cool though.

    -bs
    • by walt-sjc (145127)
      You should care because this kind of pipe (or tube) is used on backbones. A fast backbone ensures that you and 30 million others just like you can all download pr0n, browse MySpace and play WoW ath the same time at your full local link speed.
  • I wonder what operating system they used for this. NetBSD?
    • by rtyall (960518)
      OS/2 Warp-speed edition?
    • I wonder what operating system they used for this. NetBSD?NetBSD favors portability over speed, and benchmarks demonstrate this. Linux tends to win the benchmark game, and NetBSD is one of the worst BSDs in this aspect.

      Maybe they used it, but I don't see why they would.
    • I wouldn't be surprised if no OS was used at all and the bandwidth was measured by some sort of hardware network analyzer taking data directly from the cable.

      Also, this is probably raw data throughput, without any protocols as overhead on top of the payload: i.e. the entire Ethernet frame size including headers was taken as the basis for what is considered "data transfered" (- if Ethernet was even used...)
  • Technology anyone? (Score:2, Interesting)

    by Anonymous Coward
    Why do I see no post above my threshold about:
    fast photodiodes
    fast multiplexers
    GaAs-transistors
    fibre amplifiers (this is for the post about connecting continents)
    ?

    They say they do it electrically, so they need to have a photodiode with 200 GHz bandwidht,
    compare that with the diode in your DVD!
    • well, the HHI has incredibly fast photo diodes... you can buy a 100GHz PD for less than $20000 from them...
  • by Anonymous Coward
    optics.org / FibreSystems Europe reports: "Nippon Telegraph and Telephone (NTT) of Japan has demonstrated optical transmission of 14Tbps over a single fibre 160km long. The transmission consisted of 140 channels of 111Gbps each using complex DWDM techniques."
    107Gbps... pfft... yesterdays news. :)
  • Uh, this summary is missing the obligatory Library of Congresses/sec.
  • So NTT transfers 14 TB/s over a single 100 mile fibre. (=112,000 Gb/s) And Siemens manages to transfer about one thousand of this (107Gb/s). In what way is this a record, or even interesting? http://techfreep.com/ntt-sets-download-record-at-1 4-terabytes-per-second.htm [techfreep.com]
  • The Fraunhofer Institute [wikipedia.org]: now you can get your patented and licensed MP3's even faster!

    --Rob

  • I can't wait until ISPs here start offering this connection, so I'll finally be able to get a 100 Gb ps connection*
    *with 1kbps upload

"Mach was the greatest intellectual fraud in the last ten years." "What about X?" "I said `intellectual'." ;login, 9/1990

Working...