100G Ethernet

Metcalfe Pitches Terabit Ethernet

Bob Metcalfe thinks 1 Tbit/s Ethernet is inevitable, and he also believes the industry will have to tear down some standards to get there.

The "Father of Ethernet," as he's often called, will be delivering one of the keynotes at next week's OFC/NFOEC show in San Diego, sketching what he believes will be the path to get to 1 Tbit/s speeds.

And in an exclusive LRTV interview, Metcalfe points out that the path might leave behind the equipment and even the fiber that's been in the network for years.

"There comes a time when standards have to be overthrown," he says. "We're going down into sort of a dead end. I think that dead end is deep enough that we'll get to 100 Gbit/s. The evidence is, to get to terabit Ethernet, we'll have to break out of that dead end."

Metcalfe, of course, has no problem with that kind of thinking.

"There's now room to break loose of the stranglehold of standards and now move into some really fun new technologies," he says.

But would anyone even need terabit Ethernet? It's a question Metcalfe dismisses with his usual candor:

"We build it; they will come, I'm sure. It's happened every time for 35 years."

You can see the whole interview here.

— Craig Matsumoto, West Coast Editor, Light Reading

Interested in learning more on this topic? Then come to Ethernet Expo: Europe 2008, the premier event covering the hot topic of carrier Ethernet network technologies and services in Europe, the Middle East, and Africa (EMEA). To be staged in London, April 14-16, admission is free for attendees meeting our prequalification criteria. For more information, or to register, click here.

Page 1 / 2   >   >>
Pete Baldwin 12/5/2012 | 3:47:15 PM
re: Metcalfe Pitches Terabit Ethernet So, it looks like Metcalfe is going to promote a start-from-scratch agenda to get to Terabit.

I'm sure the equipment makers won't mind, but I wonder how well that will play with carriers.
IPnightly 12/5/2012 | 3:46:52 PM
re: Metcalfe Pitches Terabit Ethernet I propose an addition to Moore's Law, called Schmoore's Law, which states, "Sure transistor densities and bandwidths double every 18 months, but the usefulness remains the same." For example, I cannot download any more useful information than I could 10 years ago, because my 100 byte news item is now padded with multi-megabyte ads for cars, Scottrade, wonder drugs, etc., etc. Another example? My dual 1.8 gHz Pentium is no faster that my 400 mHz Pentium because of bloatware, security features, etc., etc. I used to run OrCAD on 640K of RAM... what can I do in 640K now? Login?
paolo.franzoi 12/5/2012 | 3:46:51 PM
re: Metcalfe Pitches Terabit Ethernet
I like to tell people if you need more than an 8051, 22V10 and a 555 - it is probably to complicated and not worth doing.

Pete Baldwin 12/5/2012 | 3:46:50 PM
re: Metcalfe Pitches Terabit Ethernet Schmoore's Law ... I like it!
lightnesmonster 12/5/2012 | 3:46:45 PM
re: Metcalfe Pitches Terabit Ethernet Seems to me that the optical components needed for serial Terabit is just the tip of the iceberg. (I assume serial otherwise it's warmed-over DWDM)

What will the Ftau of transistors need to be to serve such critical functions as modulator drivers at terabit speeds? Anybody making such parts? How are they packaged? Will they cost as much as the optics given that volume will likely be low?

Assuming one can get there, in seems that only the Infinera's of the world are in a position to address the requirements because heavy integration of both optical and electronic functionality is a necessity. As for a new fiber infrastructure; somebody must be smoking some good stuff!! Terabit over copper anyone?

The path to Terabit is paved with parasitics and too little green! Place your bets folks!

IPnightly 12/5/2012 | 3:46:41 PM
re: Metcalfe Pitches Terabit Ethernet Ahhh... the old 555 timer... you're dating yourself! I admit that my original comment may be a tad cynical... but I'd like to see some conservation in terms of bandwidth (and memory)... I really don't like full motion video ads on my text web pages. Cheers to all!
materialgirl 12/5/2012 | 3:46:40 PM
re: Metcalfe Pitches Terabit Ethernet IPnights,
you must be over 30 years old.
rjmcmahon 12/5/2012 | 3:46:40 PM
re: Metcalfe Pitches Terabit Ethernet Yeah, I don't like all these ads either. Neither did many in the public during the early days of RF broadcasting.


In March and April, 1922, Commerce Secretary Herbert Hoover sponsored a national Conference on Radio Telephony, which in part addressed the question of radio advertising. During the meeting Hoover warned that "It is inconceivable that we should allow so great a possibility for service to be drowned in advertising chatter", and the conference recommendations for advertising standards would have restricted it to near non-existence. The final report called for "toll broadcasting" to be the least important of four categories of stations, with limited transmitting ranges, and their development kept under "close observation". Moreover, commercial messages were to be "indirect" only, and "limited to a statement of the call letters of the station and of the name of the firm responsible for the matter broadcasted". The conference report, however, was never adopted as official policy, and a year later, the report of the second national conference did not include any restrictions -- or even references -- to toll broadcasting. However, the industry continued to cast a wary eye on developments, and at the third conference, in 1924, Hoover famously warned that "if the speech by the President is to be used as the meat in the sandwich of two patent medicine advertisements there will be no radio left". However, he added that "The listeners will decide in any event. Nor do I believe there is any practical method of payment from the listeners."
paolo.franzoi 12/5/2012 | 3:46:37 PM
re: Metcalfe Pitches Terabit Ethernet
So, going back to the beginning of time. Ethernet came out and fell on its face. I remember working on Ungermann-Bass gear connecting some Data General Minis together with Ethernet.

It was much simpler to use serial interfaces for our need and we abandoned Ethernet.

What really got things going was the push of PCs into the workplace. Workgroups were set up independent of IT organizations. These workgroups used Ethernet instead of Token Ring, as they had no access to the Mainframe. IT ignored these, and they spread like wildfire. Then IT realized that they could grow their budgets by taking over these networks, and the Token Ring/Ethernet wars occurred.

Mainframes became less of the focus and Ethernet won that battle after several years.

Then came 10BT, 100BT, and GigE. We have not really seen 100GigE yet and the issues of analog nature of high speed technology have yet to be addressed.

So, if they build it they will come did not work for Ethernet. Something else changed and then Ethernet became viable.

delphi 12/5/2012 | 3:46:33 PM
re: Metcalfe Pitches Terabit Ethernet Just to add a bit of history to brookseven.

It was just not the advent of the PC. the devolution of the mainframe and terminal connectivity started with the mini from DEC and others. DECNet was really Ethernet in disguise. The focus was distributed CPUs connected to central processing vs. terminals connected to central processing. Ethernet won because it became an open standard that PC manufactures could adapt and exploit at cost with margin.

Token Ring lost because of the centralized nature of the architecture.

Bob is only pushing buttons in the market as he has done so well over the years. Terabit Ethernet will come; but the physics are still years away.

It is a debate in 2012, not 2008.
Page 1 / 2   >   >>
Sign In