Slashdot Mirror
Researcher Only High Bandwidth Network
Icarus1919 writes "A brand-new 10 gigabit per second per user optical fiber network is now available to researchers in the U.S. (compared to Internet2, which offers only 10 gigabits of bandwidth total, regardless of the number of users). The National Lambda Rail, as it is known, is named for the 40 different wavelengths of light it uses to send data within the fiber network. In the past, researchers have complained about the relatively (relative when you're dealing with terabytes of data) small bandwidth they can access to send data, and the addition of the NLR will most likely be a boon to research."
From what I understand, they will be using quite a bit of the bandwidth in this as well. Do we know how much data must be trasfered at once? Is this continuous data, or is it in chunks? How much ram would it take to hold all of this data until it can be placed unto a disk for storage?
-- johntracy.com, because everybody else is wrong.
I did a paper on this for my introductory networking class LAST YEAR and the topic had been a subject in the class for a few semesters before mine.
------------------------------ SirPhreak - "It's Thinking..."
US scientists enjoy big bandwidth boost
15:21 26 October 04
NewScientist.com news service
The world's biggest fibre optic network entirely dedicated to scientific research is now in place in the US. The National Lambda Rail will allow scientists to exchange more data at faster speeds than via the internet.
"It's a landmark because it's the first time that we in the research community really own and control the underpinning infra-structure vital to advancing science," says Tom West, the CEO of NLR, based in Cypress, California, US. He will outline the benefits of NLR at the Supercomputing Conference 2004 on 6 November in Pittsburgh, US.
The days of "figuring out how to jam a lot of data into a small pipe" are numbered, explains Scott Colburn, a network engineer at the National Center for Atmospheric Research in Boulder, Colorado, US, which is part of the NLR consortium - a collection of research institutions and private companies.
"It was kind of like the first computers," he says, "except back then we were trying to work with very little memory. Now we have so much memory we don't give a hoot and in future we won't give a hoot about bandwidth."
Transfer rates
NLR is entirely owned by the US research community and offers users 10 gigabits per second each. In contrast, Internet2 is a slice of internet infrastructure currently designated to the US academic community and provides a total transfer rate of 10 gigabits per second to be shared between all its users.
Both networks use a technique called Wave Division Multiplexing (WDM) to send different wavelengths or "lambdas" of light through an optical fibre simultaneously, with no interference. Light is routed to its destination using prisms which extract particular wavelengths.
Whereas Internet2 dedicates just one lambda to the whole US research community, NLR dedicates 40. Its creators are now distributing cards embedded with lasers of different frequencies to the first four users of NLR, allowing them each to hook into their dedicated channel.
"NLR is another landmark in the progression towards ubiquitous high-speed computing, which is essential for our research," says Julian Bunn, a particle physicist at the California Institute of Technology, US.
Large Hadron Collider
He plans to use NLR - which is connected to existing high speed scientific networks across Europe - to link up with colleagues at the Large Hadron Collider in Switzerland. When the collider goes live in 2007, the network will transfer terabytes of data directly to his lab in California for analysis.
The NLR will also allow US doctors to stream high resolution video of surgical procedures in real time to remote surgeons, allowing them to monitor crucial operations.
And climatologists - who must pool measurements from sources all over the world in real time to make weather predictions - will also benefit.
But NLR is also exciting for researchers who want to experiment with the network itself. Colburn likens it to the pioneer spirit surrounding the earliest version of the internet - ARPAnet - which was purely a research endeavour.
"NLR will provide us with the infra-structure to do things that could potentially bring the network down," he explains. "Commercial internet providers can't allow us to do experiments willy-nilly because the net is mission critical."
These "destabilising" experiments could include trying out new protocols, probing how data should best be routed from one place to another and testing whether decision-making software works better when it is installed on the fringes of the network or at the core.
Researchers have been able to test some of this using a virtual software network called PlanetLab, which sits on top of the internet in a similar way to which the internet sits on top of the telephone network. But this research is subject to internet bandwidth constraints.
"NLR will
In Soviet russia, only old Koreans profit from pictures of Natalie Portman stored on Beowulf Clusters.
the addition of the NLR will most likely be a boon to research."
:)
and, of course, bittorrent
I feel so sorry for those researchers, stuck at 20Gb/s... honestly, i know terabyte databases arent that uncommon among researchers (or that common) but honestly how much of a difference is 20Gb/s to 10Gb/s, you might save half the time but to be honest when calculations can take days/weeks this transfer rate isnt going to make a huge difference... (now in my basement that might be a different story...)
drunk chemists
I think I should go back to the academe. Will they install these also to the house of the researcher?
relative when you're dealing with terabytes of data
They should be recompressing all those movies to Xvid, then they wouldnt have that bandwidth issue. I mean isn't that what you would do if you had a network dedicated to "research"? (the above post was meant to be humorous, not trollish)
sig?
most of the Lambda team was killed in that indcident at Black Mesa a few years ago....
Me: Hello, I am a researcher and I would like access to this network
Them: And what is your field of research, sir?
Me: Why, I am researching human biology and behaviour in group situations.
Them: Well that sounds very interesting. How exactly would having 10 Gbps help you.
Me: Glad you asked. First I download all the pr0n I can find, then I watch it...
Them: Look, this is about the 1000th application we have had that wants to download porn. Good day, sir.
Me: But, but....
Them: I said good day!
Chaos will always win out over order because chaos is more organized
AnimeNEXT anime convention
So Bush was right and there are internets?!?!
"There is no time, sir, at which ties do not matter," Jeeves, (Jeeves and the Impending Doom)
... oh wait, this isn't Fark. My bad.
You do deserve some credit for being the first to mention pr0n, MPAA/RIAA, Bittorrent, etc.
This is really not so different from how the present internet got started. Will researchers pave the way for a new international fiberoptic network?
When things get complex, multiply by the complex conjugate.
how does this compare to a station wagon full of DVDs hurtling down the highway?
"A brand-new 10 gigabit per second per user optical fiber network is now available"
As opposed to what - a stinkin' old 2 gigabit per second per user optical fiber network?
I haven't seen too many disk sub-systems out there that can store data faster than 640 Mbits per second. Will 10Gbit per second really be that useful? Maybe I am wrong...
10 gigabit per second per user
What is that in breasts per second?
Hi, I'm studying the effects of DVD-R isos, gimmie gimmie gimmie.
There are "Internets"!
I think the main application of this improvement is that it could boost the video conferencing applications or video streaming (which has seen so less improvement in the recent days). When talking about 10 Gb/s bandwidth, it does not help anything with our system unless you are dealing with huge amounts of data - like video streaming or conferencing. Also considering the fibre optic communication(FOC) systems it is been quite some time since we knew the potential of the FOC. Seriouly, we had 1Tb/s experimental systems 4 years back and only now do we have a 10Gb/s per user system. make love make: *** No rule to make target `love'. Stop.
I think the main application of this improvement is that it could boost the video conferencing applications or video streaming (which has seen so less improvement in the recent days). When talking about 10 Gb/s bandwidth, it does not help anything with our system unless you are dealing with huge amounts of data - like video streaming or conferencing.
Also considering the fibre optic communication(FOC) systems it is been quite some time since we knew the potential of the FOC. Seriouly, we had 1Tb/s experimental systems 4 years back and only now do we have a 10Gb/s per user system.
make love
make: *** No rule to make target `love'. Stop.
ps:Sorry I forgot the html formatting and reposting it for better understanding!
In the past, researchers have complained about the relatively (relative when you're dealing with terabytes of data) small bandwidth they can access to send data, and the addition of the NLR will most likely be a boon to research."
The complaints will continue. There's always a larger data set to move, and complaining is human nature. This new network is a good thing but are they spending time/money on educating the researchers on the most efficient use of the network? That'll increase the longevity of the network.
These posts express my own personal views, not those of my employer
... in the US?
What do you need to be researching? Who do you need to be affiliated with?
Do people like RMS count?
"Nine times out of ten, starting a fire is not the best way to solve the problem." - my wife
You're that fat security guard that got away, aren't you?
Nerd: Derogatory term typically directed at anybody with a lower Slashdot ID than you.
this only makes the Firehose bottleneck (too much data from communication channel than you can process or even store) worse...
device guys, and computer architecture gear up and give us some good processing. looks like everybody else is working!
make love
make: *** No rule to make target `love'. Stop.
Yes that is true they actually did a comparison, which if I remeber correctly endedup here on slashdot about the "bandwidth" of the US postal service just using Netflix DVD rentals and AOL disks as the "data" being transferred. It was astonishing that with just those the bandwidth was something like a factor 300 times faster than then internet in mbps and resulted in more total data being transfered than the internet over the course of a month.
Actually I came to that realization myself a few years back at the hight of my MP3 collecting days. A 40gb drive passed among friends through the mail was much faster and had better results than looking on the net.
"This is really not so different from how the present internet got started."
"The Victorian Internet by Tom Standage".
I'd say it got it's start earlier than people think.
I'm doing a research project on "how many gigabytes of porno can somebody download in one month".
> The world's biggest fibre optic network entirely > dedicated to scientific research is now in place > in the US. The National Lambda Rail will allow > scientists to exchange more data at faster > speeds than via the internet. In other words, DWDM technology has dropped in price. > "It's a landmark because its the first time > that we in the research community really own and > control the underpinning infra-structure vital > to advancing science," says Tom West, the CEO of > NLR, based in Cypress, California, US. He will > outline the benefits of NLR at the > Supercomputing Conference 2004 on 6 November in > Pittsburgh, US. I wonder why it took so long to get fiber to their local CO, then off via the ILEC to a central hub for that region? These regions then link together in some common topology we would all recognize. > The days of figuring out how to jam a lot of > data into a small pipe are numbered, explains > Scott Colburn, a network engineer at the > National Center for Atmospheric Research in > Boulder, Colorado, US, which is part of the NLR > consortium a collection of research > institutions and private companies. Yea, those were the days. But now you own the fiber and have solid leases on the rest. You simply upgrade the interfaces and backplanes every 2 years or as you can afford. > It was kind of like the first computers, he > says, except back then we were trying to work > with very little memory. Now we have so much > memory we dont give a hoot and in future we > wont give a hoot about bandwidth. I hope this type of thinking decreases with the software I use in the future. > Transfer rates > NLR is entirely owned by the US research > community and offers users 10 gigabits per > second each. In contrast, Internet2 is a slice > of internet infrastructure currently designated > to the US academic community and provides a > total transfer rate of 10 gigabits per second to > be shared between all its users. I imagine the interfaces the fiber are pluggng into run at 10 gigabits per second. One has to wonder what the backplane the SFP interface runs at. Take for example a GigE card in a juniper M40. The card, due to the backplace, can only run at 800 Mb/s. > Both networks use a technique called Wave > Division Multiplexing (WDM) to send different > wavelengths or lambdas of light through an > optical fibre simultaneously, with no > interference. Light is routed to its destination > using prisms which extract particular > wavelengths. Calling the filter a prism is quite accurate in a way. Its main goal is to cut out a certain wavelength while reflecting all the rest of the signal. They position these filters in a crafty pattern that allows the light to bounce, and all the while catching the light in collimators after passing through the filter. Picture a bank shot in pool, but some of the ball goes through the rail to travel on. > Whereas Internet2 dedicates just one lambda to > the whole US research community, NLR dedicates > 40. Its creators are now distributing cards > embedded with lasers of different frequencies to > the first four users of NLR, allowing them each > to hook into their dedicated channel. Each SFP (small form-factor pluggable) interface can be set to different wavelengths. They are all the same. It sure does sound like each location gets a wavelength. This allows easy back haul from the centralized regions. You can now have 40 diff locations in that region connect to the central hub for that region. Then yuo backhaul all 40 of those wavelengths over one pair of fiber to the main congegration point. Saving big bucks on long haul fiber. > NLR is another landmark in the progression > towards ubiquitous high-speed computing, which > is essential for our research, says Julian > Bunn, a particle physicist at the Califor
You may remember that the "internet 1" was a researcher-only network once and we all know how that turned out.
With even more competition now that the Electric companies can offer broadband service I think we'll see a Moores-law type situation in Telecom (albeit with a longer cycle then 18 months). With networks like this serving as a proving ground for new technology I think we'll see a speed-race among providers. Americans love Horsepower, RPM, GhZ, and they may not know it yet but Mb/s. Shane
That's recent history.
Why should I expect more? All I imagine is that extra bandwith will be used for more digital TV and other on demand suckdom.
Friends don't help friends install M$ junk.
There are multiple large scale IP networks, and the Internet proper isn't the only international one. I suppose it depends on how you use the term. Since an intranet is a network within your building or at most orginazation, then a research network like I2 would qualify as an internet. However there is only one Internet with a capatial I, that being the one we connect to.
Difference between the onter internets and THE Internet is more or less barrier for entry and scope. The Internet has no real barrier for entry other than the money and hardware to get a connection. It's not like I2, which requires you to be a research instution. Also the Internet has reach across the world far above and beyond any other network.
How many libraries of congress per second is that stationwagon?
Do you want your confidential data shooting across this network in plaintext?
Alternatively, do you want to encrypt data fast enough for it to be worth it?
I rarely criticize things I don't care about.
That's a great speed, but how much data can be processed on the recieving end? Would it just shove it into RAM or ? I'm just trying to think how a network card would handle/distribute that much data at once.
CB#@(*(#$_@J
free ipod and free gmail!
By "user" they mean "institution connected to our network" not "individual person". As the previous reply said, they're limited to 40 wdm channels, one per user. To put this in perspective (from wikipedia):
Anyone know what the shannon limit for single mode fiber is?
-jim
So what happens when some reasearcher jams the laser shield open with a crate, and subsequently blows a hole in the side of the Lamda research complex?
More information is available at nlr.net including a network map. The first link that went up was between Chicago and the Pittsburgh Supercomputer Center which is run by CMU (where I go). The only problem I think we only have 1 (maybe 2) gigabit links to them, so the bandwidth isn't available onto campus.
There is a weird mix of members in this. Not what I would expect from their quotes. One of the interesting ones: University of New Mexico representing the state of New Mexico. http://www.nlr.net/members.html
does this mean they are selecting light colours / frequencys to route across the physical network? you want to send to computer1 blue light routes more directly too it with less relay points.
switching paths by frequency could really speed things up especially if the frequency used represented the path between computers. and the breaks in transmission the data
Blarney Quality Restaurant, Plants
I have customers that have single databases over 20TB in size and petabytes of storage in a single datacenter, the biggest well over 25PB. They would pay millions of dollars ever month to be able to replicate that to a disaster recovery or bunker site at realtime.
http://www.leadmagnet.50megs.com
Someone else do the math - I'm tired...
RS
Shoes for Industry. Shoes for the Dead.
Really, I just want to download all of the articles. Really.
Kindness is the language which the deaf can hear and the blind can see. - Mark Twain
You know, that is an extremely interesting approach to this.
Not really feasible, except in small-ish mesh networks,
but very, very, very interesting.
Let the source of the transmission deal with selecting the wavelength; let the laws of physics deal with the actual physical routing.
It's a 40 Gbps network scaling up to 400 Gbps, it's not 10Gbps/per user, it's not 'news' because it's been in the works for more than two years and has been online (at least partially) since March of this year...who's writing this stuff and where has their head been for the last two years?
What's really unique about this network is that it is owned and managed by the research and academic partners instead of by the govt. or a service provider. It's going to allow for some really cool things to go on because not only is it going to provide service, it is going to provide a test bed for new technologies and applications related to networking that were not possible before.
http://www.nlr.net/
Yeah, they're using DWDM.
I think a network like that would be good in the commercial sector, I think I'll go lobby to make sure that it's open for public use. ;-)
We'll see if I'm right or not in 20-30 years. I don't think we'll see these kind of speeds until then.
Wait until AOL gets their hands on it.
Let the pr0n flow like WATER!
Congratulations on the understatement of the century.
First they create "The Internet" and only researchers are using it. Then it gets flooded with non-research traffic, so they create "Internet2." Now that's "flooded" and they want lambda rail. It's a cycle that's tough to end. I thought it would be cool to get us on NLR, but the futility of it was noted by my Director. Our Internet2 link (OC-12) is far from saturated, so there's really no need for it.
"Nature doesn't care how smart you are. You can still be wrong." - Richard Feynman
Sounds like we are a step closer to not needing secondary storage anymore.
We're getting this at my university! The University of Louisiana at Lafayette is on the up and up!
compared to Internet2, which offers only 10 gigabits of bandwidth total
So if I run ethernet to the machine down the hall, and do a big file transfer, it slows down the entire INTARWEB-2 for everyone?
I'll stick with INTARWEB-1, thanks.
so now I can look forward to a computer that will download the source code for the internet, compile a boot loader, that will compile the kernel, that will compile the internet.
All while it reboots and I take a piss.
God is real unless declared as int
Just wait until we see the researchers' machines turn zombie and start DOS attacking...
It seems you are making the assumption that literally every bit they transfer (and every bit you transfer, by making that assumption) will be stored on a storage device. I'd doubt that is the case ...
The Internet's nature is peer to peer - 20050301_cs_profs.pdf
Hey Slashdot -
What are the largest files moved around? From the wide-array telescope project? Genome and protein info? obligatory porn joke inserted here?
Physics is like sex: sure, it may give some practical results, but that's not why we do it.
Last week we celebrated
the 50th anniversary of CERN (European Laboratory for High Energy Physics).
There was an "Open Day" organized (visits to tens of interesting places at CERN) and the Computing Center was one of the most popular places.
I was honoured to be one of the guides there.
The LHC will produce proton-proton collisions 40 million times per second.
Even after a first pre-selection of the events, there will be around 100 "potentially interesting" events per second that have to be recorded for further (so called "off-line") analysis.
There will be 4 detectors running on LHC (ATLAS, CMS,LHCb and ALICE).
The size of the data describing a typical "event" (collision) depends on the detector, but it is of order of Megabytes.
Assuming 1MB/event/detector that gives
1MB*100events/s*4 detectors*86400s/day=
34560000 MBytes of data produced every day.
That is almost 35 Terabytes of data per day (24 hours).
We expect to have some 10 Petabytes of data per year. And we hope to run the LHC for some 20 years.
Obviously, this amount of data cannot be stored/analysed locally; it will be therefore transmitted to scientific institutes worldwide.
You may have a look at the presentation:
[PPT]
PDF
The answer to computing power is the Grid: more information at http://gridcafe.org/
cheers,
Piotr Golonka
CERN IT/CO
I wouldn't know whether to masturbate to the pr0n I'd be downloading or the transfer speeds...
Actually I was thinking the same thing when a professor showed a one minute movie demonstrating TEM tomography this morning. It was over 200 MB, so I guess he'd never heard of data compression. Actually some shorter clips are regularly available for download by students. They're also uncompressed. Is this a trend of the age of cheap storage and bandwidth or just limited to a few computer-illiterate professors?
They aren't doing any research into body odors at the moment.
...rumors on the, uh, internets.
| Ceci n'est pas une pipe.
Bet anything hosted on it would be immune to the /. effect ;)
http://www.baylor.edu/Lariat/news.php?action=story &story=21045
if you look at the picture in the story this network goes right up i-35. $200K + luck = 10Gbps
lose != loose
If you took all the data they sent in 58 seconds and put it on floppies, and then you laid all of those floppies in a row, it'd form a line 1/3 of a mile long! You could send over 1000 copies of the Complete Works of Shakespeare in that time! And if you took all of that data and wrote it in a 6 point font on a large sheet of paper which you handed out to all of the people in a state the size of Texas then the total amount of money they would have to spend to deal with the damage caused by eyestrain would be more than the entire GDP of a small Central American country.
Doesn't it make you feel good to know that our freedoms are protected by politicans, lawyers and journalists.
Isn't that a little petty?
artilery artillery
inteference interference
excercise exercise
And of course the fun ones, like tht and dosen't does not make you stupid any more then I'd seriously suggest that Bush is stupid simply because he sometimes (very publicly) acts that way.
Quack, quack.
wtf is wrong with the mods here.. seriously you guys are total idiots, I have no idea how that qualifies as a troll. Morons.
Yea... i'm like a researcher or something... uh... how can I get one.