Optical/IP Networks

Metcalfe: Sonet, Fibre Channel Are Toast

Ethernet has had a long career of bobbing, rope-a-doping, and taking down competing technologies such as Token Ring. Now that it's turning 30 years old, has the champ still got game?

Absolutely, says Ethernet co-creator Bob Metcalfe. Speaking yesterday at a 30th birthday celebration for Ethernet held at the Xerox Corp. (NYSE: XRX) Palo Alto Research Center (PARC), he said Ethernet's still got the power to go 15 rounds with anybody.

Here's how Metcalfe sees Ethernet faring against the three latest contenders -- or, in his words, the "three new Godzillas it's about to crush."
  • Fibre Channel: "It's doomed. It's dead. It's beginning to smell." Metcalfe sees IP eventually winning the SAN battle, especially once 10-Gigabit Ethernet starts to deploy.

  • Sonet: Same kind of story, with a mumbled aside about how the phone companies continue forcing this one on us. Better to inject the network with optical Ethernet wherever possible.

  • Cellular: With 802.11 ("wireless Ethernet," to many) access points popping up like dandelions, cellular doesn't have a chance when it comes to data -- although, Metcalfe conceded, the two will coexist for a while.

Metcalfe, a general partner with Polaris Venture Partners was the star of the afternoon program, held for an invite-only crowd of 300 or so Silicon Valley luminaries and media (see 3Com Salutes Ethernet at 30). He also moderated a panel of Valley vets who swapped stories about Ethernet's old days and the struggle to get the protocol accepted.

As legend has it, Ethernet was born in a 1973 memo by Metcalfe, who invented the concept with fellow PARC researcher Dave Boggs. And it's been a battle ever since. Metcalfe rattled off a few dozen challengers Ethernet has faced: other PARC protocols, technologies from other companies such as Apple Computer Inc. (Nasdaq: AAPL) or Hewlett-Packard Co. (NYSE: HPQ), and political rivals within the Institute of Electrical and Electronics Engineers Inc. (IEEE) 802.3 standardization efforts.

So, what's kept Ethernet on top? Well, it's partly a semantic trick: "Ethernet has changed and changed and changed -- and thank goodness, they still use the same word for it."

But like Doctor Who's regenerations, Ethernet has retained some core characteristics. It's simple and flexible, and that's helped its longevity.

In fact, Metcalfe said he chose the word "Ether" because he didn't want Ethernet to be associated with a particular media such as co-ax cables or copper twisted pair. That flexibility helped keep Ethernet relevant as technology advanced. "It planted the idea that [Ethernet] was going to evolve."

(By contrast, a bad word choice was "collision," used by Metcalfe in describing how traffic shares the ether. It caused trouble early on. "People know what collisions are. They're metal and glass and blood flowing everywhere. They didn't want collisions in their network.")

But the real reason for Ethernet's success wasn't so much in the technology as in the way it was sold: " 'Ethernet' is a business model," Metcalfe said.

The model was radical for its time: Ethernet was licensed for cheap, but it was enforced like a standard.

During the panel session, PARC colleague David Liddle, now general partner at U.S. Venture Partners, said Xerox charged a one-time license fee of just $1,000. That's in contrast to the huge fees associated with Token Ring.

Xerox's stipulation was that the technology couldn't be changed -- it had to interoperate with all other Ethernet implementations. "Thus we made a playing field in which we could all thrive and compete," Liddle said.

The low price helped Ethernet spread and fueled an entire industry of startups. And the continued interoperability between generations helped its foothold grow.

— Craig Matsumoto, Senior Editor, Light Reading
Page 1 / 9   >   >>
green 12/5/2012 | 12:00:53 AM
re: Metcalfe: Sonet, Fibre Channel Are Toast ethernet has a impressive list of victories. But it is time for Ethernet to morph once more to support SLA. Anyone who has used VOIP phones over the internet know that it not toll quality like circuit switched networks. that is what the newer technologies like RPR are trying to do. retain the low-cost packetized architecture of ethernet/IP while providing SLA. go RPR !!!
whyiswhy 12/5/2012 | 12:00:53 AM
re: Metcalfe: Sonet, Fibre Channel Are Toast Saying Ethernet is better than Sonet or Fibre Channel is like saying left hand threaded screws are better than right hand threaded screws, hex head is better than slot.

VC: Spin-think and pure hype gets turned into money. Amazing. But true.
gea 12/5/2012 | 12:00:53 AM
re: Metcalfe: Sonet, Fibre Channel Are Toast I'd be interested to see specifically what Metcalf said about SONET. I've seen people predicting its death for many years, and every time you ask them why it's clear they don't understand it. Metcalf sounds no different.

And of course, if 10GbE with SONET framing emerges as the standard for Metro packet networking, does this count as SONET, or Ethernet?
firstmile 12/5/2012 | 12:00:52 AM
re: Metcalfe: Sonet, Fibre Channel Are Toast It's always enjoyable to see Ethernet compared to Token Ring in the same breath as comparing Ethernet to SONET.
I am a big believer in the benefits of Ethernet, BUT, statements like that set back the entire Ethernet effort.
mbw 12/5/2012 | 12:00:51 AM
re: Metcalfe: Sonet, Fibre Channel Are Toast Metcalfe clearly does not understand the fundamental value that SONET and FC bring to their respective user environments. You don't just substitute SONET and FC with Ethernet and get these values for free... that have to be layered in. When will these types learn that trying to make 'one size fit all' kills the economics. A bus is a bus, and a train is a train, and plane is plane, and you will waist a lot of time and money bundling... ah, I won't bother explaining this again.

jim_smith 12/5/2012 | 12:00:50 AM
re: Metcalfe: Sonet, Fibre Channel Are Toast 3 motives for Metcalfe for saying what he said.

1. Ethernet is my baby.
2. I'm talking at the 30 year anniversary of Ethernet.
3. Polaris needs the hype! How else do you plan to sell these doomed startups? Huh?!
alchemy 12/5/2012 | 12:00:49 AM
re: Metcalfe: Sonet, Fibre Channel Are Toast that is what the newer technologies like RPR are trying to do. retain the low-cost packetized architecture of ethernet/IP while providing SLA.

Do the math:

Cisco OSR 4 port OC12 / 4 port Gig-E combo card $110K

Cisco 16 port Gig-E card $25K

Yeah, nobody pays retail but the price difference between quad OC12 and 16-port Gig-E is staggering.

When IEEE 802.17 becomes real and RPR ASICs are stable, Gig-E counter-rotating rings with 50 mSec protection switching are going to be very inexpensive. Unfortunately, 802.17 doesn't look like it will emerge until next year.
OneDesignGuru 12/5/2012 | 12:00:49 AM
re: Metcalfe: Sonet, Fibre Channel Are Toast He must have had so much to drink before his speech. I can name 20 reasons why SONET and FC will be around at least for another 2 decades. I cannot believe he was that naive. I have heard arguments from MetCalfe's supporters before in mid nineties regarding their love child Ethernet vs. various technologies. Every time, someone made a mockery out of them. I wonder who will do this time.
bear 12/5/2012 | 12:00:47 AM
re: Metcalfe: Sonet, Fibre Channel Are Toast Granted, he "invented" ethernet, but this guy is so far removed from the carrier world he has no idea what he talking about. Sure, future flavors of RPR *might* encroach on SONET's market share, but in no way is that happening any time soon. If I had to pay big $$ to attend this conference I'd be royally pissed about the garbage I got for top dollar.
rtfm 12/5/2012 | 12:00:45 AM
re: Metcalfe: Sonet, Fibre Channel Are Toast Bob Metcalfe is too smart to thing SONET will die from Ethernet of today. But, he really means the future flavor of ethernet, which really does keep evolving. Throw on bells and whistles (or steroids or brains) onto ethernet, is it still ethernet?

This leads to the question that will this change one reason I think Ethernet was successful - its interoperability and *backwards compatibility*. Everytime something changed/improved, people didn't have to throw gear away, and could mix and match (autosensing): 10, 100, 1000 (on copper, anyways). I know I'm being loose with terms here as to what is ethernet, and flavors, but so was Metcalfe. Will future evolutions still be backwards compatible?

Page 1 / 9   >   >>
Sign In