Forgot your password?
typodupeerror
Networking The Internet Technology

100-Petabit Internet Backbone Coming Into View 137

Posted by kdawson
from the movies-on-whim dept.
lostinbrave notes laboratory work that could lead to long-haul network cables capable of exceeding 100 Petabits per second.kilometer. "Alcatel-Lucent said that scientists at Bell Labs have set an optical transmission record that could deliver data about 10 times faster than current undersea cables, resulting in speeds of more than 100 Petabits per second.kilometer. This translates to the equivalent of about 100 million Gigabits per second.kilometer, or sending about 400 DVDs per second over 7,000 kilometers, roughly the distance between Paris and Chicago. ... The transmissions were not just faster, they were accomplished over a network whose repeaters are 20 percent farther apart than commonly maintained in such networks, which could decrease the costs of deploying such a network."
This discussion has been archived. No new comments can be posted.

100-Petabit Internet Backbone Coming Into View

Comments Filter:
  • Blu-Ray (Score:5, Funny)

    by daybot (911557) * on Tuesday September 29, 2009 @09:07AM (#29579157)

    ...or sending about 400 DVDs per second

    That's just about enough to cope with today's worldwide porn output, but what happens when the industry switches to Blu-Ray?

    • by kimvette (919543)

      I don't see why porn addicts would want to watch porn at 1080i. I mean, do you really want herpes blisters and gonorrhea drips to plague your fantasies? Seriously now!

      I don't see what you get out of porn anyway. You don't get the girl, so what's the point? Why not put that money into a decent wardrobe and haircut, and you know, actually socialize and get a girlfriend or boyfriend?

      I know, I know. This is slashdot, who am I kidding? Some slashdotters have downed a few too many bags of Cheetos and 2-liter bot

    • by von_rick (944421)

      That's just about enough to cope with today's worldwide porn output, but what happens when the industry switches to Blu-Ray?

      Folks will start getting blue balls?

  • Too bad (Score:2, Funny)

    by Spazztastic (814296)

    Too bad nobody in the USA will ever get that. Even if we were to get a connection that fast, it would have a 20GB/mo cap so the second you stream one HD flick on Netflix, your cap is filled and you're stuck at a measly 768kbit/sec down until the first of the month.

    • Re: (Score:3, Informative)

      by Idiomatick (976696)
      I really don't think this was intended for end users. You could have all media saved on computers over the course of a week. Whining that you wont get that seems extreme. Also I doubt HD vids on netflick are 20gigs.
      • I really don't think this was intended for end users. You could have all media saved on computers over the course of a week. Whining that you wont get that seems extreme. Also I doubt HD vids on netflick are 20gigs.

        I'd be happy with a fraction of a 100-Petabit connection with no cap. Most people are stuck with lousy 768/128kbit DSL or Comcast with their shared lines and bandwidth caps. Some don't even have that luxury, they have to use dial-up.

        • Re: (Score:3, Insightful)

          by thesandtiger (819476)

          I agree - the fraction I'd be happy with would be 9/10ths. Totally reasonable!

          Being serious, this is only indirectly for end users, and people bitching about slow connections here would be like me bitching in a NASA thread about how it isn't fair that NASA has crafts going 20,000 MPH while my bicycle is still stuck at a max of about 30mph. Different toys for different uses. This is clearly an infrastructure tool, one that offers much better speeds and lower costs of deployment than the current stuff.

          That sa

          • Being serious, this is only indirectly for end users, and people bitching about slow connections here would be like me bitching in a NASA thread about how it isn't fair that NASA has crafts going 20,000 MPH while my bicycle is still stuck at a max of about 30mph. Different toys for different uses. This is clearly an infrastructure tool, one that offers much better speeds and lower costs of deployment than the current stuff.

            My entire point is that even if this was deployed, the end user such as myself would probably still be capped at an unreasonable 5mbit download and a fraction of that for the upload. The USA has fallen behind severely in internet speeds while other countries are providing 100mbit right to your door at the same cost.

            That said, I'd really be happy if I could just get FIOS where I live. It is absurd to me that, living in downtown Chicago, I can't get anything better than Comcast cable.

            Up until about a year ago you couldn't get FIOS in Philadelphia. They're now starting to deploy it in some parts but it's going to take a while to be deployed. It'll be nice for people in the cit

            • by Trahloc (842734)
              It'll trickle down, slowly, but it will. The carrier hotels will be upgrading to this eventually so the USA ISP's wont have much excuse except being cheap bastards... pretty much the same as now except it'll be extremely obvious when a small ISP no one ever heard of offers 100mbps both ways to the home and comcast is still offering their measly 768k.
          • by poetmatt (793785)

            I think more accurately the point they are making is that advances in technology are showing that there is no reason for the ridiculously low usage caps we are facing.

            I don't exactly love a 250GB cap with comcast, but trust me that 50MB down/9MB up for $80 (including basic cable which is pretty bullshit because it's 90$ without) - is alright. That's what I'm getting in Evanston.

            • by iamhassi (659463)
              "I think more accurately the point they are making is that advances in technology are showing that there is no reason for the ridiculously low usage caps we are facing."

              Exactly. Anything that makes it cheaper to deliver higher speeds filters down to us. Any time AT&T, Sprint or Tata [wikipedia.org] is thinking "Gee, we'll need another 50 Tier 1 lines to keep up with demand" and scientist come along and say "No, you can do it with one 100 Petabit line, and you can use fewer repeaters", it saves them money and allow
              • by poetmatt (793785)

                absolutely agreed. Much lower latency may up server requirements a little as the traffic increases but I'll take it any day hands down.

              • I'd take an extra 9 of reliability any day over either of these things.
          • Re: (Score:3, Funny)

            by aynoknman (1071612)

            people bitching about slow connections here would be like me bitching in a NASA thread about how it isn't fair that NASA has crafts going 20,000 MPH while my bicycle is still stuck at a max of about 30mph.

            Don't know about you, but I want to go to the grocery story at 20,000 MPH, and be able to bring back a container full of stuff too!

            • by Kjella (173770)

              Don't know about you, but I want to go to the grocery story at 20,000 MPH, and be able to bring back a container full of stuff too!

              If you live so far from the grocery store you need a 20,000 MPH craft to get there, it's time to start growing your own.

              • Re: (Score:3, Funny)

                by nacturation (646836) *

                Don't know about you, but I want to go to the grocery story at 20,000 MPH, and be able to bring back a container full of stuff too!

                If you live so far from the grocery store you need a 20,000 MPH craft to get there, it's time to start growing your own.

                Do you know where one can purchase grocery store seeds?

    • Re: (Score:2, Insightful)

      by Spazztastic (814296)

      I don't see anything OT about this thread, but apparently since it doesn't have to do with the theory of deploying 100Petabit fiber, it has to be OT. It's not like I'm throwing in a hot grits Natalie Portmen comment. Mod me down more, I have plenty karma to burn while you groupthink mods waste your points.

  • second.kilometer (Score:5, Informative)

    by Paul Rose (771894) on Tuesday September 29, 2009 @09:13AM (#29579223)
    Maybe old hat to you network engineers, but I was previously unfamiliar with "bits per second.kilometer".
    From the PC World article:

    The measurement takes into account both speed and the ability to maintain it over distance, by multiplying the network's speed by its distance in kilometers. In this case, a network with an aggregate speed of 15.5T bits per second (Tbps) was able to maintain that speed over a distance of 7,000 kilometers (4,349 miles), or roughly the distance from Paris to Chicago

    • by dsginter (104154) on Tuesday September 29, 2009 @09:22AM (#29579365)

      Maybe old hat to you network engineers, but I was previously unfamiliar with "bits per second.kilometer".

      This is equivalent to 43 LoC/HI (Libraries of Congress per hour-inch).

      • by hattig (47930)

        How does that compare to cmol/LoCm (light speed moles per Library of Congress Metre).

        • Re: (Score:3, Funny)

          c mol / LoC m = (3x10^9 m/s) mol / (10 TB) m = 0.00003 mol / byte s. I'm at a loss for how to interpret this dimensional measure. Maybe it has something to do with the number of monkeys needed to type the works of Shakespeare in a specified amount of time.

        • Well lets think about this logically.

          If 43 LoC/HI = b/s.km
          and there are 14 cw/f (cubic whales per football pitch) in each cmol/CoCm
          According the google the exchange rate is 1.7.
          So a quick calculation gives us...

          about 7.
      • Re: (Score:2, Funny)

        by sharkey (16670)
        Can you convert that to rods and hogsheads please? I'm a little lost.
    • Re: (Score:3, Insightful)

      by Megane (129182)
      What I want to know is how many bits there are in the first.kilometer and the third.kilometer.
    • by jgs (245596) on Tuesday September 29, 2009 @10:06AM (#29579905)

      Maybe old hat to you network engineers, but I was previously unfamiliar with "bits per second.kilometer".

      Thanks for the info. No, this is not old hat to network engineers. I've never heard of it and I've been working in the industry for more years than I care to admit. I think it might be old hat to marketing people though, since it appears to be a classic BIG MARKETING NUMBER. Normal networking people would call 15.5 Tbps * 7000 km... 15.5 Tbps.

      Maybe it's true that optics geeks really do use numbers this way, I dunno. But the fact it comes from an AlcaLu press release doesn't lend it a whole lot of credibility.

      I am massively unimpressed by the headline on the Slashdot story. Maybe another article headlined "kdawson swallows inflated AlcaLu marketing fluff hook, line and sinker" would be in order?

      • Re: (Score:3, Interesting)

        by Rising Ape (1620461)

        Given that optical fibre capacity is limited by dispersion (different parts of the signal travelling at different speeds, causing adjacent symbols to overlap), it's a reasonable number - both a longer distance and a faster symbol rate make the problem worse. So if this is what's limiting you, you can double the distance by halving the speed, or vice versa. Of course, that's not the only limiting factor, and IIRC some forms of dispersion don't scale proportionally with distance, so it's not the only relevant

        • Re: (Score:3, Informative)

          by jgs (245596)

          At the very least I assume we can agree that "100-Petabit Internet Backbone" is a gross misrepresentation of what the press release describes. "15.5 Tbit long haul" would have been accurate.

      • Re: (Score:2, Interesting)

        by alx5000 (896642)
        I thought it was quite common to express the capacity of an optical system by its bandwidth-distance product [rp-photonics.com]... Or are we talking about something different here?
        • Re: (Score:3, Insightful)

          by jgs (245596)

          Yes, the story headline is talking about something totally different. I mean, how do YOU read "100-Petabit Internet Backbone"? Most people would not interpret it to mean "15.5 Tbps delivered over 7000 km." (The headline error is repeated in TFA. Ironically if you click all the way through to the AlcaLu press release [alcatel-lucent.com] the headline is more accurate: "Alcatel-Lucent Bell Labs announces new optical transmission record and breaks 100 Petabit per second kilometer barrier".)

          I will grant you that optics geeks ma

          • by epine (68316)

            The bandwidth distance product is useful if they're planning to compete with Seagate. The line storage capacity would increase nicely if they stretched it out to one full earth circumference, with the benefit of bringing the stored data back to its point of origin. With a 200ms rotational latency, the IOPs would suck, but the parallelism would be great with a deep request queue exploiting deterministic access times.

            I never thought of this before, but with control of a large enough bot farm, you could keep

      • Re: (Score:1, Funny)

        by Anonymous Coward

        Maybe another article headlined "kdawson swallows inflated AlcaLu marketing fluff hook, line and sinker" would be in order?

        Now that you mention it, most kdawsom stories can be adequately described by stating "kdawsom swallows". I propose a tag.

    • Re: (Score:1, Insightful)

      by Anonymous Coward

      So it is a unit that sounds really big but really isn't all that useful? I would much rather have the bandwidth between point A and point B, not the bandwidth plus the distance of the cable between the points.

      I mean, I could say the longest cable in my house are carrying signals with 105 (Gb/sec)(ft) just because I have 105 foot cable between 2 Ethernet switches running at 1Gb/s? ( (1Gb/sec)*(105 feet))

      Or would a better way of saying is that normal Cat6 1Gb/s Ethernet with a 100 meter distance limitation ru

    • by kimvette (919543)

      So why don't they just say "bits per second" without any distance spec? That just introduces a new rating that no one anywhere ever uses. I mean, I've dealt with 4 km optical fiber links, but the speed was just a mbits/second rating. Not "x megabits per y km" rating. That only confuses the issue, causing people to say "wtf does that mean?"
        Isn't it just easier to say that "the medium supports 100 Petabits per second over a 1 km link?"

      • by smoker2 (750216)

        So why don't they just say "bits per second" without any distance spec?

        Because 80% of their target market (2 people tested) wanted a really stupid metric. congratulations, they succeeded. Now you have no idea what they have achieved, but the PHB will think it's good, so you'll have to implement it next week.

    • Actually, it's quite common; I studied that at the University on my Ms. It is an effect of the signal dispersion in the fiber, which is increased with the distance. Therefore, the speed (bits per second) decreases with the distance.
  • by grumpygrodyguy (603716) on Tuesday September 29, 2009 @09:13AM (#29579227)

    I would trade this in a second for a guarantee that the last mile problem will be resolved in my lifetime.

    It's been 10 years and I'm still stuck with a crappy 1.5m/256k (1.2/180 actual) ADSL line.

    • Re: (Score:1, Funny)

      by Capsy (1644737)

      The wonders of 2mbp/s internet.

    • by i_ate_god (899684)

      While the telcos are exaggerating the scenario with too many people using up too much bandwidth, that doesn't mean it can't / won't happen. The telco's should be laying down new cables, but that infrastructure exists in their country only. At some point, that country will saturate the bandwidth of the undersea cables.

      So yeah, this is part of the overall picture for better speeds.

    • by Shrike82 (1471633)
      I feel exactly the same. Perhaps with the promise of decent underlying backbones the ISP's and hardware owners will finally shell out some money and upgrade the most critical section for residential customers. Holy shit, I made my laugh so much I spilt my cup of tea...
    • Re: (Score:2, Funny)

      by RealGrouchy (943109)

      Well, you can solve about 38% of the "last mile" problem if you switched to metric. Then you'd only have to worry about the last kilometre.

      - RG>

    • I know how you feel. My ISP advertises "Up To 3/1mbit" - which translates into roughly 2700kbit downstream (good!) and 450kbit upstream. (not so good)

  • Will we notice? (Score:3, Insightful)

    by maggotsforbreakfast (1646317) on Tuesday September 29, 2009 @09:14AM (#29579233)
    Does anyone know what percentage of our current trans-atlantic bandwidth we are using? The full article says that we currently have 10 Petabits/s*k, so this would be about a 10x increase. Thats a lot, but less then I thought.
    • Re: (Score:3, Interesting)

      by Niksko (1355055)
      I think I read that only 10% of undersea cable capacity was/is being used. Can't get you a source, but I remember reading it when I got off on a tangent after the last undersea cable damage.
      • by jhfry (829244)

        If they can do 10x without having to maintain additional fiber runs, repeaters, and endpoints... then the 10x is a huge improvement. Sure the dark fiber is there, but it's dark for a reason... it's not cost effective to light it up.

        I don't suspect that they will be deploying this soon... but if it is more cost effective than lighting up more of the existing dark fiber... then maybe they will upgrade.

  • Sorry, I'm not quite up to scratch with those new-fangled DVDs-per-second-7000kilometers. How many library of congresses per leap-year.light-year is that?

    • by Niksko (1355055)
      I believe it's equivalent to 313471681789590822345900936 Mp3s!!!!
    • Re: (Score:2, Funny)

      by Anonymous Coward
      Well, lets see: 12 parsecs per hogs-head/bushel^fortnight = 3 onions.
  • Seeing petabit internet backbones or seeing Russia from your back porch?

  • Wow, I wasn't aware there was such an extensive transcontinental cable system in 1901: http://en.wikipedia.org/wiki/File:1901_Eastern_Telegraph_cables.png [wikipedia.org]

  • by Anonymous Coward on Tuesday September 29, 2009 @09:25AM (#29579397)
    I've had 100-Petabit/decade internet at home for a while now.
    • Re: (Score:3, Interesting)

      by L4t3r4lu5 (1216702)
      112589990684262400 (100 petabits)
      10( years)
      12 (months)
      365.25 (days)
      24 (hours)
      60 (minutes)
      60 (seconds)
      ------------
      29731345.93 (bits per second)
      29034.52 (kilobits per second)
      28.35 (megabits per second)


      This new line transfers the equivalent of one decade of fully-saturated domestic ADSL2 line (24Mb) traffic every second.
      • by swillden (191260)

        112589990684262400 (100 petabits)

        Actually, 100 petabits is 1000000000000000 bits. Communications technology uses traditional power-of-10 SI units, not the power-of-2 units.

        I think you also made another mistake, not sure where. Because I get:

        100 * 10^15 / 10 / 365.25 / 24 / 60 / 60 / 10^6 = 316 Mbps.

        So, this new line transfers the equivalent of one decade of more than 13 fully-saturated ADSL2 lines' traffic every second.

        (Calculated by typing "2k100 10 15^*10/365.25/24/60/60/10 6^/p" into 'dc')

        • Because I get:
          100 * 10^15 / 10 / 365.25 / 24 / 60 / 60 / 10^6 = 316 Mbps.

          My calculation agrees with yours. I have 100 Mbps fiber at home, and I calculate it only gives 31 petabits per decade at full utilization. Luckily, my ISP imposes no limits, but I still rarely use even a percent of the available capacity (at most a few hundred GB per month out of 32 TB potentially available).

      • You're off a bit. No conversion is necessary when converting bits to... bits. You just have to divide the time.

        See the replies below.

    • by Kjella (173770)

      I've had 100-Petabit/decade internet at home for a while now.

      Google is perfect if you want it in hogsheads per forthnight, in this case:
      100 (petabits per decade) = 340.255519 megabits per second
      But just between us, if you want to brag about your connection I'd use a more common unit.

  • What exactly does bits / second.kilometer mean? Does it mean that with more kilometers, it becomes slower?
  • by Opportunist (166417) on Tuesday September 29, 2009 @09:38AM (#29579557)

    That was MPAA chairman Dan Glickman fainting and hitting the floor 'cause nobody cared enough to catch him.

  • All these advances in speed and yet consumer ISPs can't seem to offer more than 6Mbps down and 1.5Mbps up for less than $70 a month.

    If Internet bandwidth were like hard drives, we would have passed the $1/Mbps mark last year. Instead, it's still $30/Mbps.

    What are these companies doing with these multiple Tbps backbones right now if consumers are being bottle-necked on purpose?

    • A large reason for this was the pricing protection and the delaying of network neutrality that the Bush Administration gave the telecom industry. It filled the telecom industry full of hubris. The good news is that this hubris is nothing more than hot air and once network neutrality becomes law, we will see the end of price collusion and more competition. Network neutrality is the telecom company's worst fear because it means that they must upgrade their network to deal with the increased bandwidth need
    • Re: (Score:3, Informative)

      by Snowblindeye (1085701)

      All these advances in speed and yet consumer ISPs can't seem to offer more than 6Mbps down and 1.5Mbps up for less than $70 a month.

      Thats because we don't have real competition in the US, so why should they give you more for less?

      Compare this to Germany for example, were you can get 16 Mbps for as little as 15 euros/month, 50 Mbps is available and Kabel Deutschland just announced that they are going to start selling 100 Mbps starting next year [teltarif.de]. Amazing what competition will do.

    • by Ant P. (974313)

      What are these companies doing with these multiple Tbps backbones right now if consumers are being bottle-necked on purpose?

      Using them as justification to charge you $30/Mbps, of course.

  • 100M Gbit/s / 300 mln users = .3Gbit/s = 333Mbit/s = 40 Mbyte/s

  • by theJML (911853) on Tuesday September 29, 2009 @10:12AM (#29579995) Homepage

    This is the US... Can we get this in Libraries of Congress/mile?

    • No, this isn't the US, this is the Internet.
    • Re:Conversions? (Score:4, Informative)

      by Kjella (173770) on Tuesday September 29, 2009 @11:05AM (#29580721) Homepage

      I had to substitute 1 LoC for 10 terabyte myself, but according to google 100Pb/s*km = 2863278 LoC*mph. So if you give everyone in Chicago a copy of the library of congress and they drive around at 1 mph, it'll have the same bandwidth. Simple, right?

      • by Ant P. (974313)

        So if you give everyone in Chicago a copy of the library of congress and they drive around at 1 mph, it'll have the same bandwidth.

        But you could also have one LoC travelling at 3 million mph.

        This begs the question! Why are we wasting billions on planning two-year Mars expeditions for half a dozen astronauts when we could pack 1000 people into the Library of Congress and get there in hours instead?

  • I like the new metric of DVDs per second. Do you think that will catch on?
    • Yep. RIP, "bandwidth of a stationwagon".

      Though the metric actually seems to be DVD kilometers per second. Assuming about 800 DVDs per cubic foot, and 2.7 cubic meters to a station wagon doing 120km/h on the freeway, the BoaSW is approximately 933 DVD km/sec. Leaving aside whatever the article means by "400 DVDs per second.kilometer", 12PB/sec at the speed of light in fiber (~200K km/sec), divided by 4.7GB per single layer DVD gives us approximately 510 billion DVD km/sec. That's a lot of stationwagons
  • Petabit backbones still won't be enough to keep 4chan online through all those DDoSes they suffer.

  • ..but Comcast will still try to create excuses for continually increasing the cost of broadband while finding excuses to decrease and limit the bandwidth.
    • Perhaps, but let's wait and see what becomes of the network neutrality legislation. Let's see what loopholes their might be. If the wording is unambiguous, Comcast will be f*ed. They won't be able to charge higher prices for long because competition will muscle in and be able to partially use their bandwidth because throttling or queueing will be a violation of the law.
  • PETA-bit would be much better PETA's poorly chosen blog [peta.org] name. Seriously, what is wrong with those people?
  • It's traditionally described as kilometer seconds [km x sec] and not [sec x km].
  • Screw bandwidth, these days what I want fixed is latency.

    Light can travel around the earth in about 0.02 seconds, yet latency on connections is about 10 - 30 times that. Having more bandwidth might let me download yet another movie a bit faster but it won't enable certain types of application that I think are far more interesting such as distributed computing and real time collaboration (eg: games) that currently are very limited by latency.

    • by DJ Rubbie (621940)

      Not that fast. Circumference of Earth (~40 000 km) / Speed of Light in vacuum. (~300 000 km/s) = 0.133s.

      Also, speed of light in glass fiber is about 2/3 that, so it will take light 0.2 seconds to travel around the globe in a glass fiber. Don't forget most routers take the light signals they receive, convert it into electric signals so the circuits can route the packets, before converting them back to light again. At least this is the case until full optical packet switching becomes prevalent.

Pound for pound, the amoeba is the most vicious animal on earth.

Working...