Slashdot Mirror
Bell Labs Break Record With 31Tbps Via a Single 7200km Optical Fibre
Mark.JUK writes "Alcatel-Lucent's research and development division, Bell Labs, has successfully broken yet another record after it used 155 lasers (each operating at different frequencies and carrying 200Gbps of data over a 50GHz frequency grid) and an enhanced version of Wavelength Division Multiplexing (WDM) to send information at a staggering speed of 31 Terabits per second over a single 7200km long optical fibre cable. Previous experiments have been faster but only over shorter distances or by using a different type of fibre optic cable entirely."
Too bad the bandwidth cap is only 1 GB per month.
There's no -1 for "I don't get it."
1 meter vs 7200km
Then again most LAN's are not 7200 kms long :)
And I can throw a 90mph fastball... I did it before half the pitchers in the MLB, the problem is that I can only throw that fast for 1cm and not all the way to the plate.
This is probably more applicable to ISP backbones rather than LANs. Although it'd be nice if I could move digital videos from my standard machine to my media server upstairs that quickly. I ripped my entire DVD collection which took me the better part of a year to do, now whenever we buy a new movie the first thing I do is rip them so I don't end up having to do a dozen+ movies at once.
It appears you've also become immune to reading entire Slashdot summaries.
There's no -1 for "I don't get it."
They might not rip up existing infrastructure, but they might start replacing it as the old stuff starts breaking down or requires maintenance.
Not wifi, wimax, 3g, 4g, ethernet, satellite, etc.
All those tecnologies are just "last-mile" ways to bring data from this big pipes to the users. Internet is made of optical fibre.
Do you think all the big-boys are going to tear up their existing long haul fiber and undersea trunks and replace it with something new? It'll never happen. These stories pop up on /. with disturbing periodicity and I've become immune to them.
What part of the story said they needed to tear up the existing fiber, or even lay new fiber? Sure, they would need to add new gear at the terminals, but that's cheap in comparison to laying cable.
And even if they did have to lay new cable, for this kind of bandwidth I imagine they'd have already begun planning it. The more you carry, the more money arrives.
John
Wonderful! Now my porn collection will download in mere MINUTES!
...whether a special type of cable was used, or whether just fitting different transmitters and receivers at each end of the cable will do the job without the need for putting down an entirely new fibre optic cable?
Donte Alistair Anderson Roberts - hi son!
Karma: Chameleon
TFA says it is for undersea cables, not LANs.
This is probably more applicable to ISP backbones rather than LANs.
I fear the first application will be high frequency trading, with links between bourses.
For comparison, Tokyo to Honolulu is "only" 6200 km (then 3900 from Honolulu to San Francisco). Washington DC to Paris is also 6200 km. So, as far the planet earth is concerned, it's a very realistic maximum distance of interest.
Everyone knows the internet is a series of tubes.
Donte Alistair Anderson Roberts - hi son!
Karma: Chameleon
This was likely at the request of the NSA so they could download all our traffic quicker.
Circle the wagons and fire inward. Entropy increases without bounds.
What medium are you throwing it in, treacle?
Microsoft already did this back in the 90's. They got over 80Tbps via one metre long cable. So there is nothing new.
I'm glad you read the fucking summary
don't fret it, it's the new gnaa.
world was created 5 seconds before this post as it is.
The switching is so dense and so fast, that the 7200km of cable has *in flight* 146 gigabytes of information at any given time. You can back up your typical "150GB" (143GB actual) OS hard drive and user data, and be done sending it before it starts reaching the other end (if you could buffer it to send that fast, naturally). Is that some crazy shit or what?
I fear the first application will be high frequency trading, with links between bourses.
Why? This has nothing to do with lower latency.
Why? This has nothing to do with lower latency.
Indirectly, it does. Latency is affected by bandwidth usage, and the wider your pipe, the greater the chance of achieving minimum latency.
Does this beat out the station wagon loaded with 500 kgs. of optical media averaging 50 km/hr?
Units should probably be TerraBit / Sec / Km.
That was the turning point of my life--I went from negative zero to positive zero.
Wrong, this still requires amplifiers every 100km, just like today.
09 F9 11 02 9D 74 E3 5B D8 41 56 C5 63 56 88 C0
It does a bit. The higher the speed of your link the lower the clocking delay in getting out all of the bits for a transaction. Will a couple of nanoseconds matter? With HFT it just might.
I read the internet for the articles.
...was I the only one who thought "31 tablespoons of what now?"
Um... NO.. NO.. NO...
Minimum latency is the issue and that is driven by the speed of light though the fiber. The stuff you are talking about deals with the overhead amount of time to get the data on and off the fiber and that is really more about the technology being used than the minimum amount of time you can get data from A to B.
Distance plays into this automated trading in a BIG way. Having a few microseconds on your competitor can be all the difference between making a bit of cash and walking home with all the toys. Trading companies pay BIG MONEY to reduce latency. If you can move 1 block closer to the trading platform than the other guy, you can make more money than he can. So, guess what they do?
Can I get the ISA version of the adapter card anywhere?
A pox on web designers who feel that window.innerWidth == screen.availWidth
Did the test include a simulated NSA tap, to test the impact of that optical degradation?
I can get 20,000Tbps over a 500 mile long cable right now if all I send are 1's or only 0's.
Do not look at laser with remaining good eye.
Now we can get pr0n in 302976 x 170424 video at 25 fps. It will have to be uncompressed as I don't think we have anything that can compress it that fast.
now we need to go OSS in diesel cars
latency of the processing gear is far higher than the time to travel through the Transatlantic cable.
Do not look at laser with remaining good eye.
This is the first time that transoceanic cables can be made that don't need repeaters. The speed is nice, but no repeaters mean that the cable will be a lot cheaper to build and has far fewer parts that can fail. It also won't be enveloped in an electric field that attracts sharks. And finally, it becomes a lot easier to upgrade the cable later: you only need to install new equipment at either end, and don't have to worry about the repeaters being compatible with the new signalling.
What do you mean by this? Processing of the router and photonic equipment is nano to microseconds, the travel time of the photons is in milliseconds.
No, glass resistance.
Tomorrow is another day...
No it isn't. Latency through a switch is measured in microseconds.
09 F9 11 02 9D 74 E3 5B D8 41 56 C5 63 56 88 C0
No, they'll just hook this new equipment up to the existing fiber. Thats how this works you know, right?
What do you think they are testing it on? Some new 7200km fiber they cooked up yesterday? They do this with existing trunks.
Persistent Volume manager for Kubernetes - https://github.com/dwimsey/openshift-pvmanager
What, are you shoving 3's down your pipes?
There are two types of people in the world: Those who crave closure
Minimum latency is the issue
Topsy-turvy, kiddo. For timing critical systems, it's maximum operational latency that matters.
Best case is for ricers who want to impress each other. Average and median values are what most pros are concerned with - bang for the buck.
And worst case is what those running timing critical systems look at, and spend big money on improving.
Talk to me when it's 31 Tera Bytes.
- Zav - Imagine a Beowulf cluster of insensitive clods...
They don't explicitly say that there were no repeaters for this particular test, but that is strongly implied. (Sloppy reporting.) However, they do compare it to a test done recently over 10,000km with no repeaters:
I had no idea that those kinds of distances were possible without repeaters. This is, indeed, big news.
Hurrrrr no. Bandwith is how much data you can move, and latency is how fast it takes you to ping the servers. I can send you a boxtruck full of 2TB HDDs. The bandwith would be phenomenal, the latency not so much.
A repeater is different from an amplifier. A repeater receives the signal, cleans it up in the electrical domain, and retransmits it. This has to be done channel by channel so in this experiment they would need 155 of them along with the associated mux/demux WDM gear to transpond all channels. An amplifier on the other hand amplifies everything between about 1520 and 1610 nanometers, all in the optical domain. All undersea cables have amplifiers in 'festoons' which are enclosures that sit on the ocean floor.
09 F9 11 02 9D 74 E3 5B D8 41 56 C5 63 56 88 C0
From story: fibre optic cable (no repeaters)
Do you think all the big-boys are going to tear up their existing long haul fiber and undersea trunks and replace it with something new? It'll never happen.
Amps designed to work with 1 Gbps Ethernet will work with 100 Gbps Ethernet. So the theory is all you have to do is replace optics at the end to upgrade the speeds on the fiber in the middle. You don't have to "tear up" anything.
These stories pop up on /. with disturbing periodicity and I've become immune to them.
That you don't understand (and have actively worked towards an "immunity" doesn't mean they don't contain valuable information that some of us use on a regular basis. 100 Gbps was in the same "never gonna happen" camp for a while, but I'm personally "tearing up" 155 Mbps links to install 4 Tbps (44*100G).
Learn to love Alaska
It will never leave the lab.
Latency is how much time your data takes to get from A to B, not "how fast it takes you to ping the servers".
Still confused? Hint: Some packets are bigger than your standard ICMP ping.
More bandwidth == less-latent transfer, all else being the same, simply because it takes less time to transfer an entire packet of data.
Kid-proof tablet..
If I take the "single strand" mentioned in TFS very literally, then it is indeed a single strand without amplification.
(Unless you can amplify light on a single strand, without ever breaking it into two or more strands.)
Kid-proof tablet..
If I take the "single strand" mentioned in TFS very literally, then it is indeed a single strand without amplification.
(Unless you can amplify light on a single strand, without ever breaking it into two or more strands.)
Aha. I went to the Alcatel Lucent site and read The Fine Press Release where the actual truth* was published.
So there we have at least a few facts from the source. Yes, they used repeaters in a typical undersea configuration. No, they didn't say if these repeaters were the existing erbium doped optical amplifiers or if they used some other novel technology that would require either laying new cables or dredging up the old ones and splicing in new repeaters.
* If you can accept the concept of 'Truth' in a press release.
John
it is yet another chunk taken out of the ass of long haul optical fiber cable mfgs.
Lucent-Alcatel is itself a long haul optical fibre manufacturer, Alcatel having bought the Cable Division of STC when Northern Telecom (aka Nortel) bought the rest of STC in 1991. See http://en.wikipedia.org/wiki/Standard_Telephones_and_Cables
So if we run it all through the bullshit detector: Cute and fast tech, doesn't break normal any distance limits.
Thanks for digging that up.
Kid-proof tablet..
(Unless you can amplify light on a single strand ...)
Which you can - look up Raman Amplifier https://en.wikipedia.org/wiki/Optical_amplifier#Raman_amplifier
Raman amplification is quite widely used in existing fibre networks, both to increase the distance between repeaters, and to increase the bandwidth of existing fibre runs, e.g. to allow 40 Gb/sec over fibre originally laid for 10 or 2.4 Gb/sec use. By increasing the signal strength Raman amplification can reduce the effect of the dispersion which limits the minimum pulse length detectable, and hence allow higher data rates over a given fibre.
No 3's the sharp edges get stuck on things.
Do not look at laser with remaining good eye.