Bobby Johnson, Foundry Networks

"We only care about one vendor long term. And that’s Cisco."

July 23, 2003

18 Min Read
Bobby Johnson, Foundry Networks

37254.jpgBobby Johnson, president and CEO of Foundry Networks Inc. (Nasdaq: FDRY), knows success.

He founded Centillion Networks, an ATM switch startup that was bought by Bay Networks in 1995 for $140 million. At the time, this was one of the highest prices paid for a networking startup in Silicon Valley. The company had already reached profitability after only 18 months on the scene, says Johnson. Bay Networks, which later was bought by Nortel Networks Corp. (NYSE/Toronto: NT), ended up selling over $1 billion worth of gear based on the Centillion products and technology, he contends.

Foundry epitomizes Johnson's hard-driving style. He brought the Layer 2-7 Gigabit Ethernet switching company to profitability within 18 months of starting it. It later launched a blockbuster IPO conveniently timed with the peak of Wall Street's technology bubble, in late 1999. Shares of the company on its first day of trading rose 525 percent from the offering price and closed at around $166 per share. That gave the company a market capitalization of nearly $9 billion on its first day of trading.

Not bad for a guy who started out as an SNA programmer at IBM Corp. (NYSE: IBM) back in the early 1980s. During the quarter century he has been in the networking industry, he’s been involved with some of the hottest technologies of his time, including early personal computing, ATM/Token Ring switching, Gigabit Ethernet switching, and now 10-Gigabit Ethernet switching. Whenever the market has shifted, Johnson has managed to change course.

Of course, things aren't as sunny as they were back in 1999 (see Stock Watch: Foundry Networks). The company’s stock is down a tad from a high of $212. But recently the company has deftly managed its way back to profitability and growth (see Foundry Gets a Fed Boost). Meanwhile, Foundry has been able to expand its headcount, increase its revenues, and develop new products.

Johnson is the first to admit Foundry’s perseverance through tough economic times hasn’t been easy. But so far, it appears to be a survivor. While competitors like Enterasys Networks Inc. (NYSE: ETS) and Extreme Networks Inc. (Nasdaq: EXTR) continue to report losses, Foundry is reporting profits and growth. And judging from Johnson's future product plans, he has no intention of falling behind on the R&D front.

The former Black Belt karate master says he has little time for leisure activities such as golf or tennis. He’s a hard worker and expects nothing less from his employees. That’s one reason Foundry foots the bill for takeout dinners every night for its engineers.

Light Reading was lucky to get some time with this busy executive. Check out the interview and see what Johnson has to say about:

  • Bobby's Beginnings

  • The Birth of Foundry

  • Innovation Nation

  • Bobby-Style

  • Much Ado About Mucho Grande

— Interviewed by R. Scott Raynovich, US Editor, and Marguerite Reardon, Senior Editor, Light Reading

Light Reading: Before coming to Silicon Valley, you started in Research Triangle park in North Carolina. What were you doing there?

Johnson: I worked at IBM writing code. Back then I thought SNA was the end-all to end-all. But by the time I got to Silicon Valley, the PC evolution had just begun and other things were going on. I said, "Hey there’s more to life. There’s lots of different things out there."

Light Reading: So after IBM did you go right to founding a startup?>

Johnson: No, it’s a little more complicated than that. The first startup I worked for was called Convergent Technologies. I wasn't a founder but I was a senior executive. I was there from 1981 to 1987. We went from zero to $400 million in three years. Convergent was the first office desktop PC. It existed before there was an IBM PC. We had an integrated LAN, which was not Ethernet. It was a proprietary twisted pair running at 307 kbit/s. It was great for that day. No one else had anything like it.

Light Reading: Three hundred and seven kbit/s, you say? I guess it was important to get that extra 7 kbit/s in there.

Johnson: The product was called RS422 twisted pair back in those days. And every unit was self-configuring. We were an OEM company. Our main customers were Burrows, which later became Unysis, and NCR private labeling. We had $400 million in direct end-user market sales, and with the OEM overall sales were substantially higher than that. We had almost a $1 billion impact. We actually owned the corporate desktop through probably 1986 before the PC or the Mac came anywhere close to competing with us.

Light Reading: This was pre-spreadsheet, right?

Johnson: No, we had a spreadsheet application. It wasn’t Excel in those days. We had word processing, SNA, connectivity to mainframes. We did TCP/IP before it was just IP. We had a leg up over all the other systems. We had a great email back in 1985 with an integrated telephone manager on the system.

Light Reading: What happened to Convergent?

Johnson: It was acquired by Burroughs [Unisys (NYSE: UIS) later acquired Burroughs]. We couldn’t make money and they couldn’t make money, so they bought us. This was in 1987. I left Convergent in '87 right before the Burroughs acquisition.

I left and eventually became CEO of a company called TriData Co. It was a Macintosh networking, venture-backed company. That’s where I met Earl Ferguson, who was co-founder of Centillion and Foundry. I also met Tim Heffner, Foundry’s CFO, there.

Light Reading: What happened to that company?

Johnson: My chief competitor wanted to be in the Macintosh networking business more than I did, so I struck a deal that he could go raise the cash and pay off my investors and reward the employees. The agreement was if they did that, they could take the company minus me, and try to go public. Well, six months later the Macintosh market kind of got run over. So he never was able to go public. TriData merged with a company called Avatar Technologies outside of Boston.

Light Reading: So you left?

Johnson: Yeah, I went to be a vice president and general manager for two years at Network Equipment Technologies.

Light Reading: Wow, as in what has become The old ATM switch company?

Johnson: Yes, but NET back in that day was completely different. It was a very important company. We made all our money in wide-area multiplexing. We also licensed the Cisco routing code. We built and embedded routers on our multiplexers and Frame Relay switches. I took the division to profitability within one year.

Light Reading: And why did you leave there?Johnson: The company had a philosophy that it did not want to move along with the existing T1 multiplexing infrastructure, which was kind of frustrating to some of us. We couldn’t take the price down and get performance up to compete with Cisco routers of that day. So after a couple of years, I wanted to take a break.

Light Reading: How did Centillion begin?

Johnson: Earl and I got together and we decided to get into the switching arena, but we wanted to be different. Everybody and his brother was trying to build an Ethernet switch in those days. So we said the long-term switch market was ATM. The first switching application we went with was Token Ring. We built the first Token Ring/ATM switch.

We were making good traction, but we had a lot of pressure from our VCs not to push to go public. There were already some big networking companies around, like 3Com/Cabletron, Bay Networks, and Cisco. They thought the market was all locked out.

We ended up building a great product and had a lot of great traction. We started in late 1993. By April 1995 we were profitable. It had an ATM fabric. All the Token Ring guys were loving it. IBM didn’t have anything like this. We literally couldn’t make these things fast enough. Bay looked at this and saw the traction. They made us a great offer at the time. It was a record at the time – unheard of really – $140 million. Today, that may sound small, but the biggest acquisition done before then was Synernetics for $120 million. And they were a $30 million a year, six-year-old company. We were 18 months old and had only been selling product really for one quarter.

Light Reading: That’s impressive.

Johnson: Bay went on to sell about $1.5 billion or so of Centillion-derived products. After being a vice president and general manager for six months at Bay, I retired.

Light Reading: You sound like Michael Jordan. You keep retiring, but then you always seem to come back.

Johnson: Yeah, but I don’t say that I’m not coming back. I just take a rest. I took nine months off that time.

Light Reading: What do you do when you take a rest?

Johnson: My wife is from Hawaii. So we spend a lot of time in Hawaii.

Light Reading: Do you play a lot of golf?

Johnson: No, I don’t play golf. I don’t have time for that.

Light Reading: Tennis? Surfing?

Johnson: No. I mostly visit family, and I exercise. But it’s more in the martial arts area.

Light Reading: Martial arts? Are you a Black Belt?

Johnson: I was once upon a time. I lived in Okinawa from 1970 to 1973, so my high school years were spent on Okinawa. There was a lot of training available back then. Those were the good days.

Light Reading: So you left Bay after selling Centillion and then you hung out in Hawaii for a while. Then what?Johnson: I had heard about this Gigabit Ethernet stuff, but I thought it was impossible. Remember, I had just came from a company shipping the premier ATM switch. I thought, “How are they going to handle CSMACD [Carrier Sense Multiple Access Collision Detection] at a gigabit? That would be death.”

But actually Earl and I took a look at the draft standard. It wasn’t Ethernet to us. They removed the CSMACD, and they just did point-to-point full duplex, with Ethernet framing. I said, “Well, okay, if you want to call that Ethernet – but that’s just point-to-point full duplex links just like ATM, only they're using an Ethernet frame instead of an ATM cell.” So I just thought about big cells and not little cells.

Light Reading: So then you decided to build a Gigabit Ethernet switch?Johnson: Earl and I knew that Gigabit Ethernet would herald a new requirement. Up until that time, all routers were CPU/software-based. They were reasonably slow compared to any Ethernet or ATM switches – actually they were very slow. GigE and Fast Ethernet switching only scales so far. So we figured we needed routing – a new class of routing. We actually went out and hired an ASIC team and put prototypes into the CPU fast-forwarding engine. We finalized the ASIC in approximately one year. And the first Layer 3 switch or GigE switch was born.

Light Reading: How long did it take you to get to profitability with Foundry?

Johnson: We started shipping in mid 1997, and we were profitable by Q1 of 1999. So about 18 months.

Light Reading: Which competitors keep you awake at night?

Johnson: It’s not Extreme.

Light Reading: Who then?

Johnson: We only care about one vendor long term. And that’s Cisco Systems. They have the financial wherewithal, the big sales force, and they’ve got a broad product portfolio.

Light Reading: What about the 3Com/Huawei partnership? Are you concerned that they will compete with Foundry?

Johnson: For where Foundry wants to go, that’s not a threat.

Light Reading: Why do you say that?

Johnson: Huawei and 3Com don’t have anything in our class of product. Foundry is about performance and functionality.

Light Reading: What do you think about Larry Ellison’s comments that Silicon Valley is dead and that the industry is mature?

Johnson: If you go back to the early 1990s – 1993 really – only two networking firms were funded in Silicon Valley. One was Xylan, which wasn’t really in Silicon Valley but was funded. And there was Centillion Networks, which was the company I started. So people thought that in '93 Silicon Valley was over.

Back then the mentality on Sand Hill Road was: "There is no way you guys will ever go public." So they told you to go build the product and find someone to buy you. That was the strategy. By 1998 and 1999 things changed and the strategy was completely different. A lot of companies were able to go public.

So are we just in a downward funk right now? I have no idea. I’m just saying that in the early '80s there was something like 150 disk drive companies founded in Silicon Valley, and very few of them survived.

Now, you can say that we will never return to the bubble that we had in 1999 and 2000. And I would probably say that's true. I think that would actually be good for the world. But I think we will return to innovation, and I think we will return to growth.

Light Reading: So what is the next big wave of development?Johnson: Well, in the enterprise we are moving to truly an IP-only network. The number of AppleTalk and IPX networks has really diminished from what we saw in 1997. Speeds are moving up. Prices are coming down, so people are deploying Layer 3 switching in more places than originally envisioned. You also have to manage and monitor all of that.

In the service provider market you have all these transitions, but you’re also dealing with the size of the Internet route table, denial-of-service attacks, and incremental security features that you can put on top of everything.

Light Reading: What about MPLS?

Johnson: I was going to get to that. There are still a lot of unanswered questions. Will MPLS be used just as a wide-area technology, or will it be used in the metro? Will Draft Martini or VPLS be used for metro Ethernet? Then there is the rise of IPv6, predominantly outside the U.S. right now, but we’re seeing some minor penetration in the U.S. as we speak.

Light Reading: Foundry is one of these companies that straddles the line between service providers and enterprise customers. How has the split between those businesses changed?

Johnson: If we go back to the 1999 and 2000 time frame, two thirds of our revenues came from service providers. Today, 70 to 80 percent or more comes from the enterprise, so it’s a complete flip. That wasn’t because we wanted it that way. The service provider market is challenged. The enterprise has been growing into needing the speeds of Gigabit Ethernet.

Light Reading: How have you adjusted the business to change with the market?Johnson: I’ve just divided the company into two business units, which are really miniature companies. One is an enterprise business unit and the other is a service provider business unit. Over time those business units will take on different personalities. You’ll see Foundry sending its enterprise business unit to different seminars and tradeshows than its service provider business unit. Later this year, we will physically separate them into different buildings.

Light Reading: What is your long-term view of the service provider market? Do you think it will ever pick up again?

Johnson: Actually, our service provider business has stayed strong in Canada, Japan, China, Taiwan, and, to a lesser degree, Korea. Primarily, the falloff has been in Europe and North America, where the service provider market did a major shift.

In the U.S. market AOL Time Warner Inc. (NYSE: AOL) is our only major service provider customer still around. Everyone else in the U.S. market just kind of stopped spending.

Light Reading: Everyone is talking about metro Ethernet being important for service providers. But it hasn't gotten much tracton yet in the United States. Do you see this changing?

Johnson: Yes, that’s correct. We are selling metro Ethernet in Asia and Canada. But there’s no reason why metro Ethernet won’t find a home in the U.S. We were selling into metro Ethernet providers two years ago. Unfortunately for us, all those are no longer around.

Light Reading: Do you expect a service provider recovery anytime soon?

Johnson: Things have taken a step back, but I believe later this year, and into next year, things will take a natural progression forward.

Light Reading: Foundry has managed to consistently remain profitable, even when its competitors and the market have been crumbling around it. How did you do it?

Johnson: The way I would explain it is we had some tough decisions to make. We recognized that the service provider market was in a downturn early in 2001. And we shifted resources more toward the enterprise and less on service providers, both in terms of sales and R&D. We held back on some advertising efforts and reoriented them. We hired a different set of talent in some areas to focus on the enterprise. We are actually one of the few companies that is larger today in terms of headcount than we were at the end of 2000.

Light Reading: What is your headcount now?

Johnson: It’s approximately 600.

Light Reading: What was it before?

Johnson: At the end of 2000, I think we were right around 495 to 500. I’m not saying it was easy. There were some shifts and there was some belt tightening. The good thing is that the headcount is doing very well and we remained profitable.

Light Reading: There’s been so much hype around 10-Gbit/s Ethernet. When do you think anyone will start selling it?

Johnson: I’ve been selling 10-Gbit/s Ethernet for 18 months and Gigabit Ethernet for six years. Today, we have north of 100 to 150 different end users using 10-Gbit/s Ethernet in production. That’s a nice start. We have about 5,800 customers total. So the question is how do we get the next 5,000 customers using 10 Gig?

Light Reading: When do you really see it taking off?

Johnson: Realistically, Q4 of this year to Q2 of next year. So I’d say it will start happening over the next six to 12 months.

The good thing is that the price per port is coming way down on our new Mucho Grande platform. Port prices have also dropped on the new generation of line cards on our existing BigIron and NetIron. They’re now about $15,000 per port, including the optics. So that’s about 10 times what a Gigabit Ethernet Layer 3 port costs. I think that will spur demand. As an industry, we’ve dropped the price from the $40,000 to $60,000 range down to about $15,000 list price in the last year.Light Reading: You mentioned that the next-generation line cards on the existing NetIron and BigIron have dropped in cost dramatically. What’s making them so cheap?

Johnson: Better density, newer generation of technology. They also use Xenpak optics, instead of fixed optics.Light Reading: But the backplanes are still limited in terms of capacity, right?

Johnson: They still have 8 Gbit/s per slot. But realistically nobody is even using that or anywhere close to that today.

Light Reading: Your competitors point out that the new Mucho Grande platform doesn’t support any 1-Gbit/s Ethernet interfaces. Is this true?

Johnson: That is correct. The first wave of the product is 10 Gbit/s only.

We already lead the industry in density on our existing NetIrons and BigIrons. So what we’re doing is taking the existing NetIrons and BigIrons with the 10-Gbit/s Ethernet and Gigabit Ethernet interfaces, and deploying them at the edge. Then we put a Mucho Grande class machine into the core for the 10 Gbit/s. We will have Gigabit ports out at the end of the year on Mucho Grande.

Light Reading: Why did you go with that strategy?

Johnson: Well, our customers were really interested in a new core. They already had Gigabit Ethernet and 10-Gbit/s Ethernet that they could redeploy at the edge.

Light Reading: Does Mucho Grande support IPv6?

Johnson: Mucho Grande will have a set of blades at the end of this year, beginning of next year, that will support IPv6 big time. You might see us announce IPv6 on the current blades, but there is another set of blades that blow away anything anyone else is doing that has to do with IPv6. One of the big unknowns with IPv6 is the size of the route tables. Not sure what we’re going to do about that.

Light Reading: What about the 10-Gbit/s Ethernet offering from Extreme Networks?

Johnson: I’m not too familiar with what they’ve announced. I know it’s something with 20 Gbit/s per slot. We have double that.

Light Reading: Who came up with the name Mucho Grande for the new 10-Gbit/s Ethernet platform?

Johnson: I don’t know if it was me, or if it was Earl [Ferguson, co-founder of Foundry]. But it was kind of a joke. We order food every night for all of our employees. So we were sitting around the dinner table a couple years back, and someone said, “What’s beyond BigIron? We can’t just call it BiggerIron. And we can’t call it BiggestIron.”

We knew we needed a code name. “So what’s big? Mucho. How about MuchoIron? No, that doesn’t sound good – Grande Grande?” We went through a lot of this, and eventually we threw out some names like that. Eventually Mucho Grande stuck with the engineering guys.

So what was just jovial joking became Mucho Grande, the internal code name. The original Layer 2 switch and the original Layer 3 switch were called Jack Hammer and Buzz Saw.

Light Reading: So what's beyond Mucho Grande?

Johnson: I don’t know. But, indeed, we are thinking of something beyond it, but it doesn’t have a name yet.

Light Reading: What does it do?

Johnson: It always starts off as the switch fabric. So this is called the Terathon chip. It’s what is used in the Mucho Grande now. I won’t be too specific, but it will provide 100 Gbit/s or greater per slot. But these things take a while to build.

Subscribe and receive the latest news from the industry.
Join 62,000+ members. Yes it's completely free.

You May Also Like