bit-tech.net

Fujitsu boosts optical networks with ADSL tech

Fujitsu boosts optical networks with ADSL tech

Future optical networks could run ten times faster than today, thanks to Fujitsu's application of consumer-grade ADSL modem technology to high-end 10Gb/s Ethernet transceivers.

Japanses electronics giant Fujtisu has announced the development of a technology which can increase data flow in data centres tenfold, enabling 100Gb/s transmission using components originally designed for 10Gb/s.

Anywhere vast computers gather, high-speed networking can be found. In data centres, it's how each server can quickly offload data and jobs to each other; in supercomputing, it's how each individual node can communicate with other nodes at a speed that makes the system seem like a single, giant computer. Increasing these speeds is also one of the biggest challenges in computing, with companies like Intel spending billions on developing next-generation interconnection technologies to boost performance.

Fujtisu's system, however, is somewhat different. Based around a discrete multi-tone (DMT) modulation/demodulation format - the same format used for consumer-grade ADSL connectivity, whereby data is divided among multiple sub-carrier waves each running at a different level of modulation - running on a high-speed digital signal processor (DSP) the system allows components designed to transmit and receive data at 10Gb/s to operate instead at 100Gb/s - a ten-fold increase in bandwidth for very little cost.

High-end Ethernet transceivers, as used in enterprise-grade products, typically have four optical channels running at 10Gb/s each for a 40Gb/s overall throughput. Using the new system, Fujitsu claims, one of these transceivers could support 400Gb/s with just minor hardware changes - providing the extra capacity needed as the world increasingly turns to cloud systems for its data processing needs.

The system has already been proven in the lab: using off-the-shelf direct-modulation laser components specified for 10Gb/s transmission, a team at the Fujitsu R&D Centre was able to successfully transmit data to a remote system at 100Gb/s - the first time such a thing has been achieved, and the first time the DMT modem format has been applied to high-performance optical Ethernet transceivers rather than consumer-grade ADSL hardware.

The team claims that almost any 10Gb/s rated components can be used to transmit and receive at 100Gb/s, with lesser-quality components that fail to maintain linearity being detected at the point of modification and the resulting device tweaked to ensure sustained data throughput at top speeds. For enterprise customers, it opens the door to the promise of a cheap, drop-in replacement for 10Gb/s technology in the near future - and, indeed, Fujitsu has reported that it is working on the creation of an integrated DMT modem for optical Ethernet transceivers, although has not yet provided a timescale for its release.

For home users, the system promises a potential boost for the growing number of fibre broadband services: current fibre to the cabinet (FTTC) systems could enjoy a tenfold boost to the cabinet to eliminate congestion, while those lucky enough to enjoy a fibre to the home (FTTH) connection could see their own speed jump by a similar factor.

Fujitsu's research team is to present its findings at the Optical Fibre Communication Conference and National Fibre Optic Engineers Conference in the US next week.

12 Comments

Discuss in the forums Reply
Blackshark 14th March 2013, 12:22 Quote
SO if I stick one of these fibre optics in my ear - will I hear the classic 56k modem handshake!!??
Corky42 14th March 2013, 12:32 Quote
Quote:
Originally Posted by Blackshark
SO if I stick one of these fibre optics in my ear - will I hear the classic 56k modem handshake!!??

No, silly.
It's a optics cable you have to stick it in your eye.
Gareth Halfacree 14th March 2013, 12:52 Quote
Quote:
Originally Posted by Corky42
No, silly.
It's a optics cable you have to stick it in your eye.
DO NOT LOOK INTO FUJITSU WITH REMAINING EYE.
kenco_uk 14th March 2013, 13:30 Quote
THE INTERNET CAME INTO MY EYES AND OUT OF MY NOSE HOLES.

The applications for this are immense. I hope it ends up being a simple drop in replacement for all the shiny FTTC cabinets that are hanging around on street corners.
play_boy_2000 14th March 2013, 15:13 Quote
The lasers are the drop in part. You still need line cards with 100gbit ports, which I might point out, are a bit hard to come by at the moment. Unless fijitsu can have this ratified as a proper Ethernet sfp standard (along the lines of 1000base-lx, 1000base-sx etc) then it's never going to be of much use, especially if the it's not compatible with existing 40/100gbit line cards.

And for those of you who think this will speed up fttn, or even ftth anytime soon.... lol.
Gareth Halfacree 14th March 2013, 15:22 Quote
Quote:
Originally Posted by play_boy_2000
The lasers are the drop in part.
Actually, the lasers are unmodified COTS parts designed for 10Gb/s transmission. Says so in the article. Which, of course, you read before commenting. In full.
Krikkit 14th March 2013, 15:59 Quote
Quote:
Originally Posted by Gareth Halfacree
Says so in the article. Which, of course, you read before commenting. In full.

:) <3
Hustler 14th March 2013, 20:58 Quote
"For home users, the system promises a potential boost for the growing number of fibre broadband services: current fibre to the cabinet (FTTC) systems could enjoy a tenfold boost to the cabinet to eliminate congestion"


Ooooh!!!!, sounds amazing, then you realise we'd have to rely on BT to actually fit this stuff into their cabinets..so, should be ready around 2020 if we're lucky.
LightningPete 18th March 2013, 17:58 Quote
Well at least theres something that the companies have to upgrade us in 2 years time, even if the tech were available now. Although they screw us out of this tech now, at least we know our FTTC services are going to improve eventually and by a sizeable amount (even if not that described).
Well done on fujitsu for research time and effort for this
play_boy_2000 19th March 2013, 20:35 Quote
Quote:
Originally Posted by Gareth Halfacree
Actually, the lasers are unmodified COTS parts designed for 10Gb/s transmission. Says so in the article. Which, of course, you read before commenting. In full.
On this side of the pond, a drop-in part generally refers to a generic, easy to acquire part that can be used in place of a more expensive custom part. How does the Queen's English define it?
Gareth Halfacree 20th March 2013, 08:01 Quote
Quote:
Originally Posted by play_boy_2000
On this side of the pond, a drop-in part generally refers to a generic, easy to acquire part that can be used in place of a more expensive custom part. How does the Queen's English define it?
That definition does not fit in this context: a drop-in replacement is exactly what it sounds like, a replacement part that requires only that the original part be removed and does not require any redesign work. An AMD A10-5800K is a drop-in replacement for an AMD A6-5400K, for example. The lasers, however, are not being replaced - hence that definition not fitting the context.

Re-read the article and what you wrote in reply, and insert "a generic, easy to acquire part that can be used in place of a more expensive custom part," see if your reply still makes sense in the context in which it was made. I posit it does not.

The drop-in part, in this case, is the modem Fujitsu has invented: it 'drops in' to a design that has a 10Gb/s transceiver and makes it a 100Gb/s transceiver. The laser is not a 'drop in' part, in either use of the phrase, because it's not being replaced at all - it's the same laser in both designs.
play_boy_2000 20th March 2013, 15:43 Quote
In any event, my point was that fijitsu has not created a drop in replacement for an optical ethernet transceiver. No doubt that the IEEE may want to integrate some of the technology into a proper standard, but that is years away, and will require new line card designs.
Log in

You are not logged in, please login with your forum account below. If you don't already have an account please register to start contributing.



Discuss in the forums