Networking: Emergence 1979-1981
LANs and DataPBXs
7.13 Ralph Ungermann and Charlie Bass and the Founding of Ungermann-Bass
Forced out of Zilog at the end of 1978, Ungermann immediately began looking for the next great technological opportunity. His inclinations led him to communications: a field he had been in ever since 1967-1970 when he served on the staff of Art Collins, the founder of Collins Radio. Ungermann had participated in creating a time division multiplexer cabling system interconnecting much of the equipment in their state-of-the-art semiconductor factory -- a precursor of local area networking. Leaving Collins Radio for Western Digital, Ungermann managed the development of the UART (Universal Asynchronous Receiver and Transceiver) semiconductor chip; recognized as the first major communications semiconductor chip, funded by DEC and designed in large measure by Bell of DEC. He then joined Intel and created the first USART (Universal Synchronous Asynchronous Receiver and Transceiver) chip. At Zilog, he followed up with the very successful SIO chip, a two-port serial asynchronous/synchronous interface chip, an extension of his previous work. He remembers his search:
"I talked to fifty companies and to a lot of people about what was going on and where they were going. What struck me was that every computer company was working on networking technology and every one told me the same thing: 'Boy, just think, if we can control the wire, we're going to control the business.' It was clear to me that that's not what the customer wanted and that there was an opportunity there.”
Ungermann began talking to others in networking such as Metcalfe, wanting to be sure he was not deceiving himself, further perfecting his vision.
Meanwhile, for those Ungermann had hired at Zilog and left behind, it became a waiting game to see what Exxon, the parent company, would do next. Surely Exxon management must know their loyalties ran to Ungermann, not Zilog, and certainly not to Exxon. Not that many would have mourned their fate, for the self-confident swagger of Zilog that kept people working until they nearly fell asleep from exhaustion had sadly slipped into an unfamiliar world of tiring doubt that made coming to work reason to find new jobs. Ungermann recalls:
"Charlie [Bass] decided, as everybody did, that his career wasn't going to be at Zilog, so we decided to get together. We started kicking things around about what we'd do.”
Even as these conversations were going on, the new Zilog management team led by Manny Fernandez, recruited from Fairchild Semiconductor, tried to make sense of its future. To strengthen management, Fernandez raided his former employer and hired William (Bill) Carrico to be marketing manager for semiconductors. Fernandez soon asked Carrico to investigate the troubled systems group; the one Ungermann had led.
Recognizing the need to bring focus to an ambitious list of development projects, Carrico convened a series of strategy meetings in May-June 1979 attended by Bass, Estrin, Davidson, Joe Kennedy, Dave Folger and Phil Bellinger. Having just introduced their Z-8000 microprocessor, much of the discussion centered around issues such as: What is the systems business given microcomputers? Can microcomputers be architected to perform as one larger computer? Should they build a multi-user minicomputer? All questions begging answers to bring coherence to their product plans. In the sorting out of what needed to be done, and which projects should be cut, Folger suggested Ariel be used to interconnect more than just Z-80 or Z-8000 microcomputers, and be reconceptualized as an "intelligent wire." One use could be to interconnect terminals to Zilog minicomputers as a terminal multiplexer.
In July 1979, Bass resigned from Zilog and Ungermann and Bass incorporated their new company -- Ungermann-Bass (UB). Ungermann later remarked in an interview with Data Communications magazine in 1980:
"I think people do what they know. My roots go back to engineering and local area networks and I’ve felt for a long time that local networks were someday going to make it big.”
Once they had decided to focus on local area networking, Bass led an engineering team strengthened over the coming months with the additions of Davidson, Kennedy and five others from Zilog. After investigating and rejecting the use of broadband, Bass orchestrated consensus for Ethernet, influenced by rumors flying around about hush-hush talks between Xerox and DEC.
With the question of access method settled, the next question was which networking protocol software to use, TCP or XNS? Their decision to use the Z-80 microprocessor, a legacy from their Zilog days, dictated their choice. Davidson remembers:
"I looked very hard at whether to implement TCP in the original Z-80 product or XNS. But we were stuck with a very limited capability in our 64K [memory size] based Zilog Z-80 systems. I decided against the complexity of TCP, and that served us very well for several years, especially when Xerox came out with their office systems."
Ungermann, meanwhile, began what would be an exasperating search for the venture capital needed to finance UB’s growth. Ungermann recalls:
"During that time, we talked to everybody in the financial community, and there were few venturers in the venture capital community.”
Venture capitalists hesitated to invest because UB lacked a seasoned marketing/sales executive, a “third partner.” Once understood, Ungermann contacted management recruiters and soon learned of James (Jim) Jordan, vice-president of Sales of Four-Phase Systems, Inc., thought to be loose in the saddle. Jordan, a veteran of ten years with Four-Phase and with seven hundred people reporting to him, regularly received calls from headhunters eager to pitch him on their once-in-a-lifetime job opportunity. Routinely dismissing such calls, this time Jordan listened. On hearing UB intended to sell networking products he readily agreed to a meeting. Why? Jordan recalls:
"The only time we ever lost to Datapoint was when Datapoint was selling a product that they called ARC. It wasn't called ArcNet then, it was called ARC; Attached Resource Computing. It was connected by a real slow -- at that time, nobody used the term 'LAN' -- but it was basically connected by a real slow speed LAN. I spent a lot of time in product planning meetings at Four Phase and said: 'Look, this is a problem. We need to be looking at it. This is the way of the future.' But I just couldn't convince the senior management that that was the way it was going to go. So I really believed in the whole concept of LANs.”
Even though UB had little cash and a staff of only fourteen engineers and no marketing people, Jordan’s experience competing against Datapoint had informed him for the opportunity and he made the “huge decision” to join UB in January 1980. Jordan remembers:
"I was bored to death. I thought it was a good idea. I liked the idea. I just believed in it. And I liked Ralph."
With Jordan’s help, UB management completed the business plan and in February closed their initial round of financing, raising $1.5 million with a company valuation of $2.2 million. Bessemer Venture Partners, Oak Investment Partners and Venad Associates divided the investment.
That same month, UB announced their first product: the Network Interface Unit-1, or NIU-1. The NIU-1, an Ethernet terminal switch that could be configured to support from eight to thirty-two asynchronous terminals, would ship in July; making UB the first company to ship unbundled Ethernet products. (Xerox had shipped Ethernet but only as part of a system.) In addition, UB publicized its vision of “vendor independent networking,” a vision very different than Metcalfe’s and 3Com’s “personal computer networking.” Jordan reflects:
"You honestly have to give Ralph a lot of credit. He was really the strategist on it. The marketplace was really viewed as all this large installed base by multiple vendors, so you've got all this DEC, IBM, Data General and HP gear, so where's the market? Well, the market is trying to get some communication capability between this heterogeneous installed base, so that's why, if you go back in the early days, and look at all the Ungermann-Bass stuff, it talks about vendor independent networking.“
In late spring of 1980, UB received inquiries from both Bolt Beraneck and Newmann (BBN) and Sytek. Each intended to bid the NIU-1 on a major RFPQ (Request for Price Quotation) being let by MIT Lincoln Labs. Bass remembers:
"We're cooperating with both BBN and Sytek, giving them everything we could to allow them to bid our product and win this contract. And we were feeling like we were in the catbird seat, that we had both of these guys bidding our product, and we're pumping them with information."
Management’s optimism turned to frustrating disappointment in September. Again Bass:
"We got a call from BBN. They said: 'Well, bad news. We lost to Sytek.' And we said: 'Oh, too bad,' and we hung up the phone and cheered -- Only to learn that Sytek had bid their own product. Well, from that day forward they were the enemy, and they were the company to hate. Now, I think we overdid it and all this got blown out of proportion, but it probably helped motivationally."
Pliner, President of Sytek, also remembers the Lincoln Labs decision:
"Lincoln Labs wanted a high-speed, host-to-host interconnectivity product. And we said "Why don't you go buy Ungermann-Bass?", and they said, "No, we want a broadband network." We bid two options: Ungermann-Bass and our high-speed broadband network that we never had, it was going to have to be developed. We won the bid on our second option. To this day, I believe Ralph [Ungermann] still thinks that I used him.”
Every sale seemed a struggle and invariably competition with a dataPBX. Jordan explains:
"The LAN companies lost business to the dataPBX guys for a while, because you couldn't really do much more for the user than a dataPBX could. You could tell him that you could, and you had a growth path and they didn't, and all that kind of stuff, but you couldn't do that much more for him right then. And big dataPBXs, in large configurations, were $300 a port, maybe $200 a port whereas our initial pricing was something like $650 to $700 per port. For us there was a real push to get to $500 per port -- it seemed to be the magic number at that point in time."
Solving the cost problem would take time. Until then, Jordan had to find customers who would buy into the bigger vision of networking; customers willing to buy a promised future now. Jordan remembers:
"It was totally a technology pitch, because you couldn't really provide an application solution, so you'd tell them how great the technology was: 'This thing runs at ten megabits,' and 'Wow, gee whiz.' “
The fledgling LAN start-ups also lacked the substantial financial and organizational resources and enviable market presence of the data communication companies. What they did have was vision and entrepreneurial drive and those differences would prove substantial.