Hacker Newsnew | past | comments | ask | show | jobs | submitlogin
Scientists push a record 57Gbps through fiber optic lines (engadget.com)
30 points by 0x4542 on March 27, 2016 | hide | past | favorite | 16 comments


Ummm. No.

This has been done before in PAM-4 using VCSELs.

http://publications.lib.chalmers.se/records/fulltext/180262/...


You must work with a lot of optical stuff...


Wait, what does 100gbe run over?


As the article explains this is not a record for fiber in general but a record for a certain type of lower cost lasers. The headline was overly sensational.


Indeed. DWDM systems are capable of multiplexing over 17.6 Tbit/s on a single fiber. Given that most fiber cables pulled these days contain 48+ fibers each you are looking at line capacities in excess of 840 Tbit/s. Still, if you strip away the sensationalized title, its a cool bit of progress.


Most long haul fiber cables pulled these days are way more than 48, at minimum 144, 288 or most commonly 864 strand ribbon. If you're trenching 300 km of fiber along a railway the cable itself is actually a small fraction of the total project costs. At most a 5% delta between 72 and 288.


interesting how that works. I've been involved with a few long haul build outs on the African continent and 48 strand is nearly universal. My understanding from talking to project managers is "more cable length per reel = less shipping costs & less splicing"

that and utilizing only a pair with 8 wavelengths on it now means they're probably good for the next 10-15 years anyway.

probably...


In an African context with limited logistics and trucking options, absolutely. And the projects are at a bare minimum budget already. I've been involved with a few, just the subcontract for concrete vaults/splice locations runs on a shoestring budget. In many cases bringing singlemode to an inland/remote city will be a dramatic improvement no matter even if it's 12 strands back to the capitol city/cable landing station location. We're talking about replacing microwave backhauls that are 4 x STM1 capacity total, serving the entire inland/up-country area of a nation, with a single 10GbE circuit. Dramatic difference. And very basic 100 GHz grid DWDM gear can be used.


nice. I'd love to swap stories some time. I'm posting this from Malawi doing some CD/PMD testing on a new 48 core that runs the length of the country.


I'm not quite sure how to sent a direct user-to-user message, or if it's possible, so I'm doing this in a reply... I would also like to compare notes and previous experience from telecom/ISP work in developing nation environments. One of the things I think will be interesting to see in the next few years is the glut of previous-generation DWDM transport gear (Ciena, Cisco, Infinera, Adva, etc) which is being replaced by US ISPs. Equipment that can do 40 x 10 Gbps channels in the 100 GHz ITU grid is becoming obsolete and pretty much everyone is upgrading to gear that can do 100 Gbps per channel. It'll be available really cheap in the used/refurb market and totally suitable for long distance applications in Africa where newly installed dark fiber is available. I already know of a few ISPs in the Seattle area that are pretty much giving away their old DWDM stuff.


it seems there is no way to message directly. I hope you see this as well -- would love to discuss further.

aaron at aagico dot com


What about trans Atlantic cables , how many strands do they have ? And are they wide core fiber ?


Most fiber being pulled nowadays has much more than 48 strand. One of my providers is currently burying 288 strand right down the road from me.


Thanks this low speed being a record didn't make sense to me.


Low-cost 100G is four lanes/lambdas of 25G. If each lane can now run at 50G then you could get 200G with binary signaling or 400G with PAM-4.


So sensational it's embarrassing.

Also, over what distance were they able to achieve 57Gbps? I can't access the original paper.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: