back to article Boffins' bonkers fibre demo: 53 Tbps down ONE piece of glass

If you've always wanted to pump more than 50 Tbps down a data centre fibre, good news: it can be done. The bad news is that right now, it needs a fair whack of boffinry and equipment. Japan's National Institute of Information and Communications Technology has laid claim to a four-fold increase over previous fibre transport …

  1. Anonymous Coward
    Anonymous Coward

    I'll just stick to my tolken ring. Much more powerful than this elfernet stuff.

    1. TRT Silver badge

      Ah, but in terms of resilience...

      Tolkien ring has a single point of failure. You have to remember to keep it well away from mount /doom

      1. Muscleguy

        Re: Ah, but in terms of resilience...

        I'm pretty sure there isn't a fibre run of any sort across the slopes of Ngauruhoe. State Highway 1 does run across the high altitude desert to the East of the volcanoes but I'm pretty sure that is just copper wire strung on poles.

        NZ's optical fibre backbone is still the pipe NZ Rail laid beside the North Island Main Trunk rail line which dives down and to the west SW of Ruapehu. It is vulnerable to lahars from that volcano such as the one which caused the Tangiwai disaster. But Ruapehu is not Mt Doom. Ngauruhoe is to the north, the smallest cone of the three.

        The volcano you really need to watch out for is much less obvious. It's the large circular lake to the north of the Central Three, the Taupo caldera. Think Yellowstone in terms of potential size. A Taupo eruption would almost certainly cut the cable.

        1. Robert D Bank

          Re: Ah, but in terms of resilience...

          if the Taupo caldera blows then fibre cable will be the least of your concerns!

          The Taupo Eruption was the most violent eruption known in the world in the last 5000 years. It is possible that ash from this eruption was the cause of red sunsets recorded by the Romans and Chinese at that time.

  2. Anonymous Coward
    Anonymous Coward

    The key question

    So when can I sign up for streaming 8k VR porn?

  3. TDog

    Sounds nice; being part of the bit BT left behind (i.e. Junction box outside, everyone within 100m radius gets SFA but the rest of Writtle gets FTC broadband) I suppose it is too much to request that we have a class XXXX* plate tectonics master who could do me a small favour. I'd be grateful for at least 14 femtoseconds or agree to take a contract on bits of BT management. (You choose 131 and I'll think about which one).

    Or I would pay a small honorarium for any enthusiastical lady or gentleman. Merely a competent plate tectonics maneuverer; or any sparse Hermione Granger with a bit of spare block time.

    1. Elmer Phud

      I'd just be happy if BT was allowed to use the fibre that runs past my house instead of being forced to go O/H

  4. Winkypop Silver badge
    Devil

    Don't tell Malcolm Turnbull

    He is still working on his Twine 1.0 tech.

  5. Kevin Pollock

    Not that far from reality

    As the headline suggests the method used here is pretty exotic, it's true.

    But if the application is Data Centre Interconnect (DCI) then there are far easier ways to do this.

    The DCI market generally needs reach of less than 150km, because it involves interconnecting a distributed "metro" data centre (one where the operator has leased buildings in different parts of a metro area and needs to interconnect them as though they were one big building). The big ICPs like Google, Facebook, Microsoft, Apple, Amazon and Softlayer do this all the time. And they have a limitless demand for capacity...seriously...check out some of their keynote presentations.

    There's commercially shipping technology that can put over 27Tb/s on existing fibre (not this multi-core stuff, which is still experimental). This commercial technology uses the C-Band. By simply extending it to the L-Band you can double the capacity to a little more than these boffins achieved. And L-Band is easy-peasy. We've known how to do it for almost 20 years, but people have tended to stick to C-Band because it's the sweet spot. Next gen technology could increase the Baud rate, or the modulaiton order, and easily double or maybe quadruple the capacity. Check out this blog wot I wrote on two post deadline ECOC papers that use actual fibre that exists today. One talks about higher Baud rates, and the other 1024QAM modulation with a technique called Constellation Shaping.

    https://www.infinera.com/infinera-makes-terabit-waves-ecoc-2017/

    But if people really do need the capacity L-Band is there. And so is the O-Band, E-Band and S-Band (in most modern fibre...older fiber may have some water peaks in their attenuation chart).

    Since the reach is so short we don't need in-line amplification on these links, and so all of these wavelengths are potentially "open for business" in the DCI market. And the key point is that you don't need exotic new fibre. This is good old SMF, LEAF, or whatever you've got in the ground today. And generaslly speaking there's shedloads of fibre around is you need it.

    So I'm pretty scepitcal of the multi-core and photonic bandgap fibre experiments. Maybe one day...but there's so much capacity still to come in existing fibre (as long as you don't need to go a long way), we're OK thanks.

    By the way...we don't make fibre, so I've no axe to grind there. If people do get this multi-core fibre to work (and can make enough of it, and can build an entire connector and splicing ecosystem to support it, and are willing to pay to lay it in the ground) then I'm sure DWDM vendors will happily use it.

    1. Jellied Eel Silver badge

      Re: Not that far from reality

      To me, the clever bit is increasing the length of multi-core fibres, ie precision pulling silicon taffy. I'm less convinced about the applications, given you'd need to buy into an entirely new ecosystem. And the actual benefits aren't clear, ie 53Tbps via 19 cores is only 2.8Tbps/core, or 7.5Tbps with 7. That's less impressive than the headline figure.

      So it'll be the TCO argument vs standard SMF in convenient 144f or 288f cables.. Which would include SLAs and MTTR to recover from backhoe fade. Restoration times on standard fibre are pretty well known, registering & splicing multicore fibres, less so. There's also plenty of data to support traditional SMF pulls, but very little regarding multicore fibre aging or stability in a typical MAN environment. There, the problem's often capacity per duct or sub-duct +/- wayleave and headaches with roads being stopped for new construction or upgrade works.

      It could be fun in other applications though, ie automotive or aircraft to reduce cable harness size or weight.

  6. Mark M.

    All nice but...

    The smutserver you're streaming your pron from is serviced by a PoS ISP with a single 10Mbps connection shared with about a gazillion other smutsurfers.

    Having a 10's of terabytes/sec download link is only worthwhile if the rest of the network is capable of the same kinds of speeds, otherwise you are throttled to the speed of the slowest link in the connection (usually at the other end of the chain or when your connection gets routed though the local intelligence gathering service).

    1. Elmer Phud

      Re: All nice but...

      Let's cut out the middle-man.

      Why doesn't GCHQ become an ISP promising breathtaking speeds?

      The small print would be interesting.

      1. Anonymous Coward
        Anonymous Coward

        Re: All nice but...

        Why doesn't GCHQ become an ISP promising breathtaking speeds

        Rather than an ISP, they'd do better to become the globe's fastest pornslinging host and bitcoin exchange. They'd still get 90% of the UK traffic they're currently scooping, but as WorldPornCentral, they'd have every ne'er do well on the planet coming along to see what they could get from the honeypot, and be able to know everything about the bad guys.

        Of course, this leads me to the conclusion the TLAs will have reached many years ago, which is that the way forward for them was/is as a dark net host. Unfortunately, like Enigma and all back doors, you have to actively overlook a lot of useful intel to avoid compromising your position, so potentially the TLA's are actively supporting a lot of low and mid level drug and violent crime, in the hope of accessing higher value intel.

  7. Pen-y-gors

    We can dream

    I just checked the Openreach website to see if this is available locally, and it reports that it is in planning and will be available sometime between now and December 2365

    1. Anonymous Coward
      Anonymous Coward

      Re: We can dream

      Unless the government subsidises it for £1trillion and then they will be able to provide it to 58% of the population by 2362.

    2. Anonymous Coward
      Anonymous Coward

      Re: We can dream

      Actually, BT/Openreach will say it should be available to you in the next 2 months, then quietly keep shifting that date by 2 months, every 2 months until 2365.

      These BT vultures have a habit of hovering over their (copper) carcass, to prevent any sniff of competition.

  8. anothercynic Silver badge

    Ooooooooo!

    I'm in luuuurve! This is exciting!

  9. Roland6 Silver badge

    Headline factually incorrect!

    Sorry but "multi-core fibres" is not "ONE piece of glass"

    The only real question is whether the technology will scale and so support the distances seen in undersea cables.

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon

Other stories you might like