MAINE--
I recently conducted an interview with the Bob Metcalfe, the inventor
of Ethernet in 1973 (with D.R. Boggs) and founder of
3Com in
1979. During Metcalfe's 10 years at 3Com (the 80's) the company
dominated LAN markets. In the 90's, 3Com ($17 billion market cap) has
been eclipsed in the networking market by
Cisco ($200 billion), but
continues to innovate under CEO Eric Benhamou with its fostering of
the super-popular Palm devices. Metcalfe remains an iconoclastic icon
whose view of the Internet that he helped build is both contrarian and
optimistic. For the last few years, he has been a primary contributor
to
InfoWorld.
3Com
During the early 70's how did being at Xerox help you? What was it like at PARC?
It was heaven at PARC. It was heaven. We had the first personal computers - I'm sorry, what we think of as the first personal computers, I don't want to get into that argument- and we had a LAN connected to them. We had an advantage we had PCs and a LAN. Our competitors UC Irvine had minicomputers and a LAN. And we had the ARPAnet. We had a gateway to the ARPAnet. We had our own internet protocol. We were in the thick of it. When Cerf started his seminar at Stanford in 1973. '73 is the big year in internetworking. That when both Ethernet and TCP/IP were invented. ethernet in may TCP/IP in summer. That was a huge year. TCP/IP didn't get deployed until 1983 when the Internet cut over from the ARPAnet.
What motivated you to start 3Com? Who did you found it with?
I started in June of '79 all by myself and attracted, began to gather people shortly thereafter; I think employee number two later that year that was Greg Shaw who had worked with me at Xerox. The goal in starting the company was to pursue the business opportunity in the consortium which I had helped form between DEC, Intel, and Xerox to pursue an Ethernet standard. The role of 3Com was to develop products that would serve that Ethernet plug compatible marketplace. Or more generally the theory of the company was that the pace of change was paced by incompatibility. That incompatibility was keeping people from connecting things and so the company's business was to connect things using standards hence the name Computer Communication Compatibility: 3Com. The three standards we started with were Unix, TCP/IP, and Ethernet. Two of three of those worked out, namely TCP/IP and Ethernet. Unix didn't quite take off the way we thought it would.
In '79 Xerox donated Altos and Ethernet equipment to Stanford, Carnegie Mellon....
MIT also.. That's right. Where was that going?
I'm just wondering about how the usage of Ethernet propogated. You had 3Com getting companies to build Ethernet equipment?
We got this consortium to make a standard. That was the key. And then we started shipping porducts in 81 conforming to that standard-although the standard wasn't really finished until 82. In 82 a magical thing happened to 3Com. We succeeded in making a Ethernet card for Intel personal computer. That made 3Com.
Is that still their bread and butter?
I think the Ethernet cards from 3Com are now a quarter of the business, I'm guessing. And modems are another quarter and they have all sorts of other equipment. I'm not an expert any more as I left 9 years ago.
Bridge Networks
In 1987 3Com made a play to get into the enterprise market by
buying Cisco's biggest competitor, Bridge
Networks. How directly were you involved in the acquisition
of Bridge Networks?
About as directly as you can get.
Why did you acquire them? What were the Bridge founders Judy Estrin and William Carrico like?
They're still alive.
At the time, I mean.
You're forever hearing about consolidation in one industry or another it seems that industries are consolidating all the time so we and Bridge decided to merge as part of a wave of consolidations that was going on there we didn't have a significant enterprise business that is with bridges and routers and stuff, we had some but not much and they didn't have a LAN business per se. So we had the idea to join these companies. It was horrible horrible.
Why was it horrible?
We did everything wrong. I think 3Com was two or three times bigger than Bridge at that time but we treated it as a merger of equals which was stupid.
Were there East Coast/West Coast problems?
They were both Silicon Valley companies. It was just stupid because we approached it as a merger of equals so we tried to be nice to everybody. We ended up with two heads of sales, two heads of marketing, two heads of France, two heads of Germany, two heads of engineering, and they spent the next couple of years trying to kill each other.
Why did they leave?
Kraus and I intended that Carrico would become the chief executive
of the combined company. Both of us at that time were riding pretty
high and planning to back out of the bus. We were tried to reduce our
commitments and Carrico looked like a good guy to take it over. When we
merged with them I stopped being chairman, Kraus kicked himself up to
chairman, Carrico was kicked up to president, and that was all going
forward. Over the next year or so Kraus and Carrico... well, there are
two sides to every story. Our side of the story is that it became
clear that Carrico did not have the right skills to run a large
company which was what it had become. He was going up a factor of four
and Bridge had been rapidly growing from there. So Kraus and the board
and everyone including me said "Oh my God! Carrico is a meddling small
company kind of a guy who doesn't know how to manage managers of
managers of managers." Therein conflict arose and Carrico left. It's
funny Estrin had maintained all along that the fact she and Bill were
married was irrelevant, that she was a separate entity. Of course she
resigned the same day, thereby completely undercutting that
theory. She left the same day Bill did.
Judy Estrin was once quoted that 3Com had the chance to be the market leader in internetworking instead of Cisco. Why didn't 3Com become Cisco?
3Com had chance to be Cisco, we had the chance to be Bay
Networks/Synoptics, we had the chance to be Novell... Same way
IBM
let DEC
get started and DEC let Sun
get started and DEC let Apple
get started. Had to do with focus and culture and competency. 3Com's
focus was on personal computers and the networking thereof and so we
didn't get into the enterprise business.
And because the Bridge acquisition just didn't work in getting into the enterprise market like you hoped?
That's exactly right. It was likely Bridge was already having
trouble with Cisco when we bought them but we didn't realize
it. Bridge was not the vehicle. Excuse me, let me add that the merger
was so bloody it's also respectable to believe that maybe Bridge had
the potential to beat Cisco but 3Com spoiled it. I'm sure that's the
Bridge point of view.
Was 3Com's 1998 US Robotics acquisition a replay of the Bridge merger?
Absolutely. Yeah, Eric Benhamou, 3Com's CEO, broke one of his rules. He had learned. Eric
came from Bridge, and had suffered and learned from this
merger of equals. Eric broke a rule, his own rule, and sort of
badly. It looks like they recovered from it but it set them
back.
Running 3Com
What was it like running 3Com? Was the market just in businesses?
It was all business. There was no residential component of the business.
What were the day to day and yearly worries?
Those days 3Com was embroiled with, was obseessed with Novell. The
competition there was over network operating systems: Novell, Microsoft,
3Com. You could argue that was one of the contributing factors of 3Com
not doing well against Cisco, because we were obsessed with Microsoft
and Novell.
How did that play out?
The network operating system of 3Com was never more than 10% of
revenue, but we wrapped 3Com around the axle of fighting Novell which
was 80% network operating system. Novell won. We dallied with
Microsoft and their LAN manager which was the competitor to Novell.
After we lost to novell we thought we could team up with Microsoft and
IBM and to finally get Novell in the end. In the process of that
Microsoft divorced from IBM and screwed 3Com. which explains some of
my lingering bitterness about Microsoft. While we were doing that that
Cisco was ramping.
Was that simply a bad choice? I know it's hard to answer contrary hypotheticals, but do you regret that?
Considering the outcome it would have been better if we hadn't been
in there. If we had succeeded it would have been a lot better...who
knows. I believe there were execution problems rather than it being a
bad idea to begin with.
What lessons did you learn from running 3Com?
You could argue I didn't really run the company. I wasn't chief
executive. I was a division manager and marketing VP and head of
engineering at various times. I was only president and CEO in the
early days. I ran the hardware division which was about 70% of our
revenues, I became the head of marketing my last year there. The first
time I ran marketing was the successful time. I ran sales and marketing
the first couple of years 81-83. The second time I ran marketing while
being screwed by Microsoft. It was very unpleasant. I had to be
bragging about aour relationship with Microsoft while they were
screwing us. I think being Microsoft's partner is a very dangerous
thing to be. We were in good company. IBM, 2Com, a long list. The key
lesson I learned at that time is what you get obsessed with. We lost a
lot of energy with getting obsessed with Novell. The right thing is to
be obsessed with customers not your competitors. That's where
Microsoft had gone wrong and why they're having trouble in
antitrust. So much of what they've done has been focused on being
anticompetitive instead of serving customers.
What did you and 3Com do right?
It did everything right. We did everything The company was built on
a very solid base. It was not a company intended to go public three
weeks after inception. 3Com's old compared to these new companies. We
were raised during the day when we were going to be the next HP,
raised for the long-term with a culture. We had to everything
right. Manufacturing was one of the keys to 3Com's success, thanks to
Howard Charney who is now at Cisco. It's world class, a great success,
accountable for a lot of profits today. Manufacturing really runs
well.
Will 3Com going to be around 20 years down the road?
I think it's going to get bought in the next few years. I'm just
praying it's going to be bought by somebody cool.
Who would that be?
Lucent would be cool. Alcatel would be horrible.
Is that because they're French?
That does contribute to it, yes. There are some companies you admire and some you can't just stand.
How about Nortel? I'm interested because they just bought Bay Networks.
Nortel would be a good one, but they bought Bay. That would tend to
delay their purchase. They'd have to wait for Bay to crash and burn
and then go buy another one. I suspect it might be bought by a Qwest
type company, one of these new things coming up. No-one knows what
they're like. What is this company that bought US West?
The Net Will Win, Although It's Junky
Pardon me if you feel I'm being glib, but do you still think that
the Internet is going to collapse in 6 months?
The Internet collapses every day. If you look at my columns on the subject I'm talking about outages. The Internet has outages of differing sizes every day, I call the big collapses megalapses or gigalapses. I'm famous for predicting a gigalapse in 1996 which didn't happen. In 96 we only got a 118 megalapse: AOL on August 7. Their Cisco routers died and they had a hard time figuring out why. I was wrong about that. I'm pretty sure a gigalapse will occur. I'm generally credited for is predicting that the Internet will go away which is not what I ever intended to predicted.
Are people simply misinterpreting you?
I'm sure you can find a sentence in my column. Unfortunately they're online all the time. You'll see that I'm complaining about the fragility of the Net and the potential for a collapse. At the WWW conference in Toronto recently I predicted there'll likely be a gigalapse in the next 10 years. I think the Internet's been lucky in not having a bigger one yet.
The whole Quality of Service question internet/phone network dropped packets/ outage?
There are two versions of that debate: one is how junky computer
equipment is compared to phone equipment, fundamental reliability, hot
standbys and reliability mechanisms and the such. The other is Quality
of Service that's a technical difference. Internet runs under a best
effort regime and doesn't agree to carry packets reliably. The
carrying of voice telephony is possible but not ideal on the Internet
for now. It's important to remember that there are two issues. The
first is carrier class reliability, the otherone has to do with this
Quality of Service notion. The Quality of Service issue is big because
if the Intenet solves it the Internet will carry voice telephony and
the telphone network will go away.
It seems like there are a lot of people trying to make that happen.
Yes, and they will succeed.
Can you say that the Intenet will win, or is it better to just say that there will be one big network?
I run an annual conference called VORTEX. It's about convergence
of the networking, phone, television industries. The theory of that
case is, going back to your statement, that the Intenet will win. The
theory is that there will be convergence but it will not be like a
merger of equals. The Internet will attract voice traffic and the
voice traffic will be almost incidental. Some of the wags there have
said that it will even be free which is terrifying to the $200 billion
a year telephone companies.
On the other hand do you see Lucent, coming from a
telecommunications background, beating the top datacom player, Cisco,
at the Internet game?
Lucent is a surprising company. It's supposed to be a big old
lethargic elephant but it's not. Lucent is not obeying the principle
that big old companies roll over and die and let fresh young companies
knock them over. And Lucent is really pissed at Cisco. They're really
annoyed at their market cap and rhetoric. They're in a fight to the
death and it's hard to pick a winner.
Always On
What excites you about networking companywise or technologywise today?
Gosh!
For example, gigabit Ethernet, which seems like it's been trying to work for a few years now?
Gigabit Ethernet is inevitable and already runs at 10 in the lab. It's not exciting, just inevitable progress, sort of automatic. The exciting thing to me is getting to the all optical Internet And that means all optical all the way out to the desk or into the living room not this DSL stuff.
How about wireless?
I'm not big on wireless. I stay away from wireless. I've been screwed to many times by wireless people making promises. I'm just waiting for them to do something. I let other people write about wireless. All optical is the way to go: always on, tremendous bandwidth, high reliability. Wireless is stopgap except for the most mobile applications. If you can run a wire you might as well. Wireless is being used as a bypass because of the regulatory problems with copper.
Wireless is just succeeding because governmental issues?
Yeah, it's artificial. I'm anti wireless. I think fibers work so well you'd never want to use wireless when you can run a fiber. It's exciting getting to the all optical Internet. We're deploying DSL, cable modems (TCMs). We're doing that full speed and that gives us always on high-speed Internet access and that's exciting but they're both stopgaps..they both using copper instead of fiber. What excites me is getting on to doing the all-optical Internet instead of all this stopgap.
Isn't Internet 2 all optical?
It is, but it's not what is about. It is faster, but the hard part of all optical is the last mile. The backbones are almost all optical now, they just need optical switches, which are coming.
So the action is in the last mile, in the home?
Yes. Not just in the home, also getting into the home. But you'll
have this high-speed always on Internet, instead of this silly dialup
we have today which is slow and intermittently connected. If we got
Amazon and EBay and Yahoo with dialup can you imagine what we're going
to get with if it's always connected at high speed?
If the Internet is always on, then everything will be on the Internet.
There's a billion applications we haven't thought of yet. Noone
thought of Ebay. There'll be EBay-class ideas that'll arrive with the
always on high-speed Internet. that'll begin with cable modems and
DSLs, but it'll really take off with all-optical.
--Adam Brate, Sociotech Correspondent
August 13, 1999
