backtop


Print 62 comment(s) - last by icanhascpu.. on Feb 6 at 2:47 PM

New device could soon usher in the 1 Tbps Ethernet era

We are seeing significant gains in the performance of many components in computers today. CPUs get faster with each new generation of the hardware. RAM gets faster, hard drives get faster, and SSDs get faster as well. What is also needed are performance improvements in the networks computers send and receive data across.

A new breakthrough is now promising some impressive gains in the speed at which we can read optical data. Today phone calls and cable TV shows are sent across buried cables using sliced light. The slicing of these signals is typically done via an electro-optic modulator. This is the device allowing the laser beam to be turned on and off at high speeds, enabling the ones and zeros of binary code to be transmitted.

One of the big challenges, and bottlenecks in today's system, is the reading of that data on the receiving end. A new record has been set in reading this type of digital data with an error-free component that was able to read data at 640 Gbits/sec.

The feat was engineered and performed by a group of researchers from Denmark and Australia and the results of the breakthrough were published in the journal Optics Express. According to the researchers, conventional readers in use today require a photo-detector to operate. These detectors are capable of operating at 40 Gbps, a mere fraction the new sensor is capable at operating at.

Multiplexing is often used to speed up data transmission with each signal being transmitted down an optical fiber at the same time. At the receiving end of these streams, the data has to be de-multiplexed to get usable data out of each of the streams sent.

The basis of the new technology comes from an experiment performed in Japan in 1998 where researchers were able to send data at 640 Gbps and read the data back, but the reading apparatus required special long lengths of cable made from a special fiber. The system was also very unstable.

The new de-multiplexing device being demonstrated by the research team at the Technical University of Denmark can handle the extremely high data rate in a stable manner. The new system also requires a waveguide of only 5 cm long, compared to the 50 meters of special cabling needed in the Japanese experiment.

The 5cm long waveguide was developed by the Centre for Ultrahigh Bandwidth Devices for Optical Systems or CUDOS in Australia. The compact size of the new device also makes it easy to integrate into other components to design ultra-fast functional chips. The researchers say that the process could also allow for even higher data transmission rates running into the trillion bits per second range.

The team of researchers says that in the near future they hope to enable a 1 Tbps Ethernet network. Compared to some of the fastest 1 Gbps Ethernet networks in us today that is a huge improvement in data transmission speed. The compact design of the required components should also mean that when they come to be, 1 Tbps networking components could be the same basic size as the network components in use today around the world.



Comments     Threshold


This article is over a month old, voting and posting comments is disabled

Time frame/Cost?
By adrift02 on 2/4/2009 11:13:18 AM , Rating: 4
While this is great and all, I am wondering what the estimated time frame is for something like this to actually be implemented. For example, if a high-speed provider (Comcast or whoever) wanted to implement this upgrade, would they need to go into each area and install new "waveguides"? Are these buried and/or numerous? I just know that if this would take going into each neighborhood and digging up lines, even with a huge speed boost and cheap "waveguide" cost (no idea what that actually is hah), they have very little incentive to get it done (see Oligopoly).

If anyone knows what this sort of upgrade requires, please fill us in...assuming it was ready to go today of course




RE: Time frame/Cost?
By Alpha4 on 2/4/2009 12:07:30 PM , Rating: 5
quote:
While this is great and all, I am wondering what the estimated time frame is for something like this to actually be implemented.
As with all things reported on Dailytech, this technology won't roll out until:
a) 15 years from now
b) when it's obsolete
c) when you've forgotten all about it

Take your pick! ;)


RE: Time frame/Cost?
By SandmanWN on 2/4/2009 1:54:14 PM , Rating: 2
yeah and this tech is also dependent on other things such as the switch fabric on the core router being able to handle that much data throughput between 50-60 cards.

I think the main benefit is power. Our current core router takes 40 OC192 blades to reach its peak of 5.4 Terabytes with a power usage of over 9000 Watts. Doing that with 6 cards and a chassis 1/4 the size would be incredible for our Co-Lo energy bills.

Although given a single core setup now is a multi-million dollar deal on top a yearly million dollar contract for upkeep, I can imagine this is going to be a huge investment that dwarfs current costs. I suppose its a good thing its probably years out from deployment given the current financial atmosphere.


RE: Time frame/Cost?
By SandmanWN on 2/4/2009 2:04:59 PM , Rating: 2
bit of an embarrassing slip-up. 5.4 terabits not bytes.


RE: Time frame/Cost?
By Jedi2155 on 2/4/2009 5:07:01 PM , Rating: 4
Since it was consuming over 9000 watts of power, all is forgiven.


RE: Time frame/Cost?
By Alpha4 on 2/4/2009 3:17:08 PM , Rating: 2
I don't proclaim to be an expert on the subject, but I don't see how this tech would necessarily have to impact any current infrastructures. If the technology was successfully rushed to market then the only concern I foresee would be manufacturers phasing out or dropping support for existing equipment a little bit sooner.


RE: Time frame/Cost?
By 9nails on 2/4/2009 10:23:12 PM , Rating: 2
I'm no expert either, but I have some insight as I work near this field. This would be sold in pairs. So the major users of this would be the backbone ISP's. Think of the really big guys on the Internet that sell bandwidth from major city to city, or possibly between continents. In some rare cases, perhaps Google or Akamai as example, would use this inside of their data centers.

They won't necessarily have to fork-lift replace anything. And they could likely re-deploy the old equipment for lower tier customer technologies and set this besides what they already own. Buy using the same fiber cable plant they own now, they could realize 16X more performance just replacing the lights on each end of each fiber. Or add 16 more customers on a line that would normally only carry 1. Profit is the reason.


RE: Time frame/Cost?
By threepac3 on 2/4/2009 2:33:19 PM , Rating: 4
d) all of the above


RE: Time frame/Cost?
By kattanna on 2/4/2009 12:15:09 PM , Rating: 5
like i stated in another thread here, this is a backbone tech. not something you are going to see rolled out to your house.

the equipment costs for routers of this nature are easily going to be costing in the 5 digit range considering 10Gb switches already cost $10,000.

and even if comcast did install this to your house and gave you a 100Mb download, you know they would still cap you at 250GB or some shit and then charge you for going over.. and/or throttle you back to dial-up speeds. ;>)


RE: Time frame/Cost?
By Moishe on 2/4/2009 2:37:35 PM , Rating: 4
Even just getting 100Mb would be "freakin' sweet"


RE: Time frame/Cost?
By 9nails on 2/4/2009 10:51:37 PM , Rating: 2
That's pretty cheap for a 10Gb switch! I recall not to long ago when they were $40K for few ports in a switch.

Our providers (Comcast, Cox, etc.) will have to replace copper cables with fiber before we see improvements in the home. Especially if we're going to move from "HD" TV signals to "Super Hi-Vision" broadcasting in 2015. (Ok, this probably isn't for the home, but don't get mad at me for my dreams!)


RE: Time frame/Cost?
By michaelklachko on 2/4/2009 1:59:39 PM , Rating: 4
No need to dig anything out. Those new waveguides are needed on the routers. They would only need to replace the equipment, not the fiber.


RE: Time frame/Cost?
By Moishe on 2/4/2009 2:34:55 PM , Rating: 2
Don't get ahead of yourself there! We aren't even operating at 100Mbps much less at 1Gb.

Granted if the backbones can be upgraded then the internet as a whole could handle more traffic.

None of this matters all that much when we are barely capable of even getting a decent broadband connection to our houses. It's a shame.


BANG!
By Procurion on 2/4/2009 10:38:27 AM , Rating: 2
When that hits my little old computer that has HDD's who read/write in the mb range, there's gonna be a smoking hole where my computer sat!




RE: BANG!
By dani31 on 2/4/2009 11:31:41 AM , Rating: 3
Not really. What you'll notice will be something like:
- more consistent bandwidth in peak traffic hours
- cheaper subscription for the same bandwith
- higher bandwith (i.e. 1.5x) for the same money you pay today.

No one will stick this device in your computer, don't worry it's designed for ISPs and public infrastructure.


RE: BANG!
By afkrotch on 2/4/2009 1:54:57 PM , Rating: 2
No one will stick this into their computer....yet. It'll eventually happen when a like 500 Tbps connection is standard for your coffee maker and a 500 Pbps connection for your computer.


RE: BANG!
By FITCamaro on 2/4/2009 2:31:13 PM , Rating: 3
quote:
cheaper subscription for the same bandwith


That made me laugh.


RE: BANG!
By Alpha4 on 2/4/2009 5:34:53 PM , Rating: 3
I feel sorry for anybody who believes an ISP would drop prices for anything, whatsoever.


RE: BANG!
By martinrichards23 on 2/5/2009 4:38:40 AM , Rating: 3
How much does your internet cost today?

How much 5 years ago?

How much 10 years ago?

I'm getting roughly 40 times the bandwidth at 1/10th the price, I'd say that prices have steadily been reduced.


RE: BANG!
By FITCamaro on 2/5/2009 7:34:08 AM , Rating: 3
today = 8 Mbps connection for $45/month
5 years ago = 8 Mbps connection for $45/month
10 years ago = 56K connection for $30/month

Obviously from dialup to broadband there was an increase but since then no.

Even in Florida speeds are 10-12 Mbps for $45/month. Unless you have FiOS, you're getting roughly the same thing as you did 5 years ago with the same price. And you might have a bandwidth cap on it.


RE: BANG!
By 7Enigma on 2/6/2009 7:14:54 AM , Rating: 1
Well you just shot your own argument in the foot. 5 years ago if the price was the same you were paying significantly more for the same service than you are now. That pesky inflation and all...


RE: BANG!
By Diesel Donkey on 2/5/2009 3:10:28 AM , Rating: 2
If you actually had that kind of data throughput available to you, why not just get rid of the HDD's and have everything stored on a server somewhere else?


Lets get...
By Screwballl on 2/4/2009 1:22:55 PM , Rating: 2
Lets get Megabit home internet connections up and going first. It is sad when the fastest FIOS connection still only runs around 50Mbps, my home network can run at 100-1000 Mbps. Heck my current ISP only has up to 20Mbps cable connection available in my area, and I subscribe to the 8Mbps service because the rest are too expensive.

Lets get cheap Megabit home connections first before jumping into Terabit LAN connections.




RE: Lets get...
By afkrotch on 2/4/2009 1:49:02 PM , Rating: 2
Ya, no kidding there. I'm currently using a T-Mobile Germany 16 meg connection. It also comes with a phone service. Costs me around 53 euro a month.

Before that I was stuck with Allied Telesis 10 meg service for $75. Major ripoff company there, but when you're stuck on a military base, you have zero competition. It was them or them. Some were lucky enough to live near the fenceline and get wireless service off the Japanese market. Service was horrid too. Slow speeds, constant disconnects, and customer service that denies they have a problem.

Pissed me off, as they were getting their bandwith off NTT for cheap. Hell, NTT service is 3000 yen for a 100 meg fiber line and voip. If you couldn't get fiber, you can just get the 100 meg vdsl line with voip.

Stateside. Fck, your lucky if you can even find a 10 meg service. Most places that offer broadband your at 3-5 meg max. Guess better than dial-up.

My parents have a 2 meg wireless cable setup. Bad weather, no internet though. Course better than the crap dial-up they had before, which cost just as much as the wireless cable.

I don't have high hopes for those living out in the boondocks to ever get broadband in some form. Too costly to bother.


RE: Lets get...
By StevoLincolnite on 2/4/2009 11:33:21 PM , Rating: 2
I'm in Australia, Paying $110 for ADSL 2 and my sync speeds are 7mb down, 1mb up, with a total of 40gb of downloads a month. (15gb on peak, 25gb off-peak data), Add in the line Rental which is $19 a month... Well you see the issue, not to mention once I hit my download cap, I'm reduced to 64kbps, or I can pay $6 per gigabyte. (Which is cheap compared to Telstra's 15 cents per megabyte!)

I'm fairly lucky though, that's a pretty good deal for me considering I am limited to the Telstra DSLAM, and that my ISP (Westnet) Started to wholesale there ADSL 2+ services, and thankfully uploads are not counted in the download limit.

The state of Broadand in most countries is pretty poor right now though, everything is getting more demanding, I find that my old 1.5mb connection that I had been using for years and years is no longer adequate for my needs while streaming ABC iView, Playing on Xbox Live! And it's not like the cost of supplying the services has decreased much over the years either, 1.5mb still costs $80 a month here for 60gb of downloads, I suppose it's better than the $120 they charged for it a few years ago, but it's still poor value.

Hopefully the Government catches wind of this technology before choosing the bidder to build Australia's National Broadband Network, and I hope PIPE manages to create more large links to other countries to bring the costs of data down.


RE: Lets get...
By FITCamaro on 2/5/2009 7:35:45 AM , Rating: 2
You know any time I think of living in Australia this irks me.


RE: Lets get...
By JKflipflop98 on 2/5/2009 1:41:40 PM , Rating: 2
I would have moved to Australia years ago if it wasnt for the crappy infrastructure.


RE: Lets get...
By MrPickins on 2/4/2009 2:20:38 PM , Rating: 2
This technology will allow the backbones of the networks to handle much more throughput. This will trickle down to the consumer in the form of faster or cheaper service.


RE: Lets get...
By shiznit on 2/5/2009 3:59:52 AM , Rating: 2
Good heavens man, I would kill someone for FIOS, the best in my area is 3.0/.768 mbit Verizon DSL for $32/month. Comcast can't serve me because of zoning laws in the county and the local cable company roughly 2.5/.256 which should be a criminal offense for $40/month.


1 Terabyte/s download speeds?
By MrPoletski on 2/4/09, Rating: 0
RE: 1 Terabyte/s download speeds?
By mmntech on 2/4/2009 11:25:03 AM , Rating: 5
I'm thinking Ultra High Definition pr0n. At 4320p, you can almost smell the shame!


RE: 1 Terabyte/s download speeds?
By anonymo on 2/4/2009 1:24:34 PM , Rating: 5
I can smell the shame at 1080p, at 4320p I would be the shame.


By GodisanAtheist on 2/4/2009 3:26:24 PM , Rating: 2
I think at 4320p, I would feel ashamed.


Still very impressive
By Shig on 2/4/2009 3:25:52 PM , Rating: 2
The fact that this is even possible impresses me.

But I also remember reading an article a while back about Australia having a 4G wireless network already that can hit speeds of the average home cable connection now.

I can't think of a reasonable application for that level of bandwidth other than high end cyber security (NSA) and the military and maybe uplinking supercomputers to one another to share simulation data in real time, but I don't think even 1tbps is fast enough for that.

+1 to getting the average home user connection much higher / affordable.




RE: Still very impressive
By Icelight on 2/4/2009 4:32:54 PM , Rating: 2
quote:
and maybe uplinking supercomputers to one another to share simulation data in real time, but I don't think even 1tbps is fast enough for that.


Even if it was fast enough (let's just assume for all purposes it is) then the key issue would still be latency. Even internal memory with, say, 400 cycles of latency ruins any performance you can get when you start talking in the hundreds of GFLOPS (nevermind T/PFLOPS!).


RE: Still very impressive
By FITCamaro on 2/5/2009 7:37:14 AM , Rating: 2
TCIP/IP protocols are severely out of date. Just by rewriting those we could see a performance improvement. But its doubtful that would ever happen since it would break a lot of other stuff.


super interesting...
By flexy on 2/4/2009 10:41:26 AM , Rating: 2
This could have significant impact on many areas like internet, digital communication in general, TV standards, compression technologies...

G.




RE: super interesting...
By icanhascpu on 2/4/2009 5:07:19 PM , Rating: 2
The speed of digital optical lines will impact digital media distribution?

My lord what an epiphany.


By someguy743 on 2/5/2009 9:16:47 AM , Rating: 1
Wow. This could be HUGE. It looks like the day where everyone can communicate via video if they want to is on the way. Kind of like in that movie "Back to the Future". They'll probably figure out a way to put this technology into Apple iPhones before long. That will be a seriously powerful iPhone. Ultra fast internet, crystal clear high definition real time video and audio in Dolby surround sound. Super cheap bandwidth.

It'll save a lot of jet fuel for people traveling on business. Why spend big bucks on traveling to a city for a sales call when you can talk to your prospect via HD video like you're in the same room with them?

There might even be cheap, super realistic holograms (like in the movie "Star Wars") so you look like you are there in the flesh for some meeting on the other side of the world. That's what massive bandwidth and superfast computing power will do.

Check out this hologram they had on CNN during the election last year:

http://www.youtube.com/watch?v=thOxW19vsTg




By icanhascpu on 2/6/2009 2:42:21 PM , Rating: 2
The ridiculous rubbish and out-of-touchiness in this post is astounding.

Its 2009. Not 1989.


Planned Obsolescence
By grandpope on 2/5/2009 11:35:58 AM , Rating: 2
quote:
developed by the Centre for Ultrahigh Bandwidth Devices for Optical Systems or CUDOS in Australia.

Seems to me these folks have painted themselves in a corner with their choice of a name. They should have chosen one more vague so they can continue to shift goals, much like the March of Dimes does.




The point is no digging
By heulenwolf on 2/5/2009 11:47:56 AM , Rating: 1
The point of this technology is that it increases speeds 16x using existing fiber optic cables. It requires no or little digging. Digging trenches to install additional lines or running miles upon miles of a whole new type of cable through existing trenches is extremely slow, risky, and expensive. Technology such as this would allow faster internet backbone, faster provider backbone, and yes, potentially faster access to your neighborhood with a relatively minor infrastructure upgrade when compared to digging up half the country. Since we're nowhere near done digging up half the country in the US to get sufficient fiber installed, it will be a long time before we all see the benefit of this technology. The point is that governments and telcos know that the fiber infrastructure they're installing and maintaining now will maintain its usefulness even as speed demand increases.




BACKBONE!!
By HrilL on 2/5/2009 5:39:34 PM , Rating: 1
Everyone's talking about getting faster speeds to your homes. Well It all starts at the backbones! and this is backbone tech. Once the backbones had a lot of extra bandwidth then it trickles down business and home connections while those cost tend to stay the same as the speeds go up over time. Some ISP are adding more limitations but on average the speeds are going up slowly for the same price. This tech will most likely be used for connections from country to country and then to the national network level and work its way down.




Ridiculous...
By quiksilvr on 2/4/09, Rating: -1
RE: Ridiculous...
By kattanna on 2/4/2009 10:45:04 AM , Rating: 2
this is a backbone technology. something the ISPs would use internally and externally to other ISP's. not to your house or inside your home as i bet you or I couldnt even afford the equipment when it does become readily available.


RE: Ridiculous...
By Yames on 2/4/2009 10:48:06 AM , Rating: 3
How is moving to the next step ridiculous? These connections are not for single computers, and definitely not intended for home use. My work is currently installing 10Gbs connections on our long haul core. This is the next step.


RE: Ridiculous...
By quiksilvr on 2/4/2009 11:08:34 AM , Rating: 1
I meant the SPEED is RIDICULOUS. Sorry didn't specify that...


RE: Ridiculous...
By Ratinator on 2/4/2009 1:35:20 PM , Rating: 2
Bandwidth has generally always been a limiting factor in how well websites perform, especially hi traffic ones. Something NetFlix will benefit greatly from something like this especially as subscriber usage increase and HD signals become the default signal. Sending out an HD signal to tens of thousands of homes will need this kind of bandwidth.

What this can also mean is that current technologies will become cheaper meaning companies currently delivering on lower speeds can move up to the next level with literally no change in price.


RE: Ridiculous...
By shiznit on 2/5/2009 4:04:21 AM , Rating: 3
no, the speed is LUDICROUS :)


RE: Ridiculous...
By tastyratz on 2/4/2009 10:54:56 AM , Rating: 2
good point.

Who would actually NEED more than 20mb of hard drive space anyways? That kind of jump is just ridiculous.

The article discusses ethernet networks as if this is supposed to be for home networks, but last time I checked an optical demultiplexor wasn't built into my linksys router. Most likely this will just mean faster internet speeds at lower prices.

For reference ddr3 1600 is 12.5 GB/s, your off a decimal ;-)


RE: Ridiculous...
By afkrotch on 2/4/2009 1:59:33 PM , Rating: 2
And the hard drive DDR3 pulls from or pushes too is around 100 MB/s average (velociraptor average fyi).


RE: Ridiculous...
By invidious on 2/4/2009 11:12:28 AM , Rating: 1
This would definately be a practical upgrade. You seem to be implying that the method of transmitting data would be faster than the data can be read from the source or stored at the destination, this is not the case. As others have pointed out this is an infastructer upgrade that will trickle down. We are no where near internet connections outrunning hard drives, and with SSDs making leaps and bounds that day is being pushed further back.


RE: Ridiculous...
By MrPoletski on 2/4/2009 11:13:02 AM , Rating: 1
Think about it...

If you can grab any information your PC might need off a remote internet location faster than you can lift it off your own mechanical hard disk..

Then why have a hard disk? why have storage at your physical location at all? after all online they have petobytes of storage available.

For good or bad, I think cloud computing is the future. You won't have applications installed on your PC, you will rent them on a per-use basis. Sounds horrible, but by this point most applciations will be internet-merged anyway and so wouldn't work properly offline.


RE: Ridiculous...
By Master Kenobi (blog) on 2/4/2009 11:40:23 AM , Rating: 1
Dream on. Data security and privacy will crush cloud computing where it stands. Notice google's web based office programs seem to be gaining zero traction?


RE: Ridiculous...
By m1ldslide1 on 2/4/2009 12:10:55 PM , Rating: 2
quote:
Dream on. Data security and privacy will crush cloud computing where it stands. Notice google's web based office programs seem to be gaining zero traction?


I disagree. Not just because this is contrary to everything I've heard around the industry, but because there is simply too much money to be made by google and others to NOT solve the security problems. Not to mention that there is simply too much money to be saved by every home and business on the planet with respect to energy consumption, hardware costs, support costs, etc. It's naive and short-sighted to think we'll continue having these 250W heaters under every desk, and multi-million dollar data centers in every medium+ sized business, and the full-time staff to maintain it all. At some point we will have simply a monitor with a wireless uplink and a high-speed network backbone providing transport - its just a matter of time.


RE: Ridiculous...
By melvin121 on 2/4/2009 12:30:06 PM , Rating: 2
funny when you put it like that it almost sound like mainframe computing...

But I still dont see your point being the be all end all. There are still people that travel and would need offline access to files/apps etc. Also your reference to 250W heaters, when they are making strides towards lower power consuming processors, and the likes of SSD where power consumption there is decent.

I think cloud computing will have its place perhaps in certain types of businesses (medical comes to mind).


RE: Ridiculous...
By FITCamaro on 2/5/2009 7:38:05 AM , Rating: 2
Maybe you want that, but I sure as hell don't.


RE: Ridiculous...
By icanhascpu on 2/6/2009 2:44:58 PM , Rating: 2
Security problems can't be "solved".

Thats like saying scientists are going to "solve" death.


RE: Ridiculous...
By zsdersw on 2/4/2009 1:03:11 PM , Rating: 2
Cloud computing may ultimately fail, but not because of "data security" and "privacy". The cat has so *totally* been out of the bag on those two things since the Internet began.


RE: Ridiculous...
By michaelklachko on 2/4/2009 2:10:22 PM , Rating: 2
It's funny that you are worried about privacy and security, even though most people in the world store some of their most sensitive data not on their home computers, but "in the cloud". I'm talking about web-based email, such as gmail, yahoo, or hotmail.


RE: Ridiculous...
By icanhascpu on 2/6/2009 2:47:08 PM , Rating: 2
Only if they were retarded.

Real backups go on tape or now secure solid state drives in a vault. Not a server room.


RE: Ridiculous...
By Procurion on 2/5/2009 9:12:24 AM , Rating: 2
Exactly! No matter how much money you throw at solving sec/priv issues, there are ten times the number of idiots wanting to hack and exploit such a thing. There is no way that I will ever put anything of value on someone elses' computer. To do so is the same as putting your desk on the curb, available to any passerby who is able to open the drawers and dig around in it.

Security? Does Google or anyone give specific information about how they secure data? You have no idea what measures they take to insure privacy; you must "trust" a multimillion dollar company to act in "your" best interests. Are you kidding? They'll dump anyone like old fish if it suits them.


"I f***ing cannot play Halo 2 multiplayer. I cannot do it." -- Bungie Technical Lead Chris Butcher











botimage
Copyright 2014 DailyTech LLC. - RSS Feed | Advertise | About Us | Ethics | FAQ | Terms, Conditions & Privacy Information | Kristopher Kubicki