Networking: LANs 1983 – 1986
LANs Over Data PBX
10.9 Concord Data Systems
Ken Miller knew he had a serious competitor when Ungermann-Bass (UB) acquired Amdax Corporation on January 1, 1983. Miller had worked closely with Amdax management to bring into being a token bus LAN standard. Miller remembers when the token bus technical specification was finalized for voting in 1983:
"We succeeded in getting token bus passed, so great, holy shit, a little company pushed that through. We pretty much wrote the standard. The other company that participated was Amdax, another token bus start-up."
Token bus was now inevitable. Concord Data Systems (CDS) no longer had to wait for a standard to be voted on before starting product development, confident that the standard would not be changed before being finalized. Soon the problem became missed engineering schedules and commitments. Creating a token bus controller proved more difficult than projected and, by mid-year, CDS still had no saleable product. Miller recalls that maybe:
"we did not keep the pressure on the LAN engineers with our focus on getting a standard passed. In any case, the product wasn't finished. It kept having delays. The RF modem was a problem because we couldn't get people. Then to do even a terminal server, we said: 'Hey we had to do the upper layer, so we thought it's going OSI.' You couldn't buy OSI layers so we ended up doing our own transport layer, which was a whole shit load of work! We were naive and cocky."
Miller's earlier opinion that local area networking was a data communication problem, not a computer problem, seemed open to question. In any case, CDS found token bus to be much more challenging than dial-up modems, the other category of products they were innovating and selling.
As 1983 progressed, it became obvious that they had to improve the performance and reduce the costs of their token bus implementation. The solution required VLSI semiconductor technology to reduce the functionality embodied in dozens of chips and, hence, printed circuit board space, into one or two chips. CDS began interviewing semiconductor foundries to both design and fabricate the chips. Management visited Intel, Advanced Micro Devices, Signetics, Western Digital, and AMI, a division of Gould. Not surprisingly, given Miller’s long association with Gould, they selected AMI. Once their custom chips became available, hopefully in a year, management believed their products would be superior to those of their competitors. They assigned Phinney the responsibility to manage the project out of Phoenix.
The more immediate need, however, was generating revenues and cash. CDS had adopted an OEM sales strategy using manufacturers representatives. An important advantage of this strategy was only a few large customers would have to be supported, not many small ones. A disadvantage, however, was that CDS had little, if any, contact with the ultimate buyers. Competing for the limited number of known OEM customers were: UB (with Amdax), Interactive Systems (a start-up now part of 3M), and even Sytek. When Phillips became their first OEM, confidence surged and an intensified effort was made to convince General Electric (GE) to switch from Ethernet to token bus. Then in mid-1983, they received a call from General Motors (GM). Miller recalls:
"They said they had this MAP [Manufacturing Automation Protocol] project. They had been talking to Interactive Systems. They had been talking to Allen Bradley about this. They said: 'Hey, you guys are furthest along. We want token bus. We've already selected 802.4 to be the lower two layers of MAP. We want to tie this thing together in the factory. So we want this multi-vendor demonstration to happen at the NCC [National Computer Conference] show in the summer of 1984.'
Well we couldn't pass that up. Here's the big corporate presence."
Miller remembers that "we hadn't finished our terminal server product," and to satisfy GM, "we diverted resources." But they rationalized: GM was worth it. If they could just pull this off, they would be uncontested market leaders.
Fortunately, Miller did not have to wonder about their modem business, for it basically kept getting better and better. When they introduced their new V.22bis 2400 bps modem, it proved to be all Miller had promised:
“In 1983, we came out with an auto-dial version, and it's really the auto-dial version that's like a Classic Coke product."
As a consequence of their increasing market presence, Roger Evans of Micom visited Miller expressing interest in OEMing dial-up modems from CDS. Miller remembers Evans wondering: "Why are you trying to build a LAN business?" Miller turned down Evans' query to buy modems, just as he had earlier declined Bass's overture to OEM RF modems.
In late 1983, CDS finally completed development of its token bus controller, the TIM, or Token Interface Module. (See Exhibit 10. Concord Data Systems Token Bus Controller) It connected factory floor devices to token bus through the V.35 interface. Unfortunately, it did not support the RS-232 interface, and, therefore, could not be used as a terminal server. CDS also implemented the lower layers of the MAP protocols in the TIM as specified in the first draft of the proposed MAP standards.
Exhibit 10. Concord Data Systems Token Bus Controller
Not having a token bus terminal server proved a problem, a serious problem. Customers wanted to buy terminal servers, not the V.35 interface controller CDS was selling. It wasn’t that CDS didn’t know they needed a terminal server for one was in development. It was just that their limited engineering resources had been diverted to supporting GM for NCC.
With their token bus TIM generating minimal revenues, let alone profits, Miller had little choice but to raise more capital. Some argued CDS should go public, after all LAN companies were hot and CDS had a convincing story to tell: they were the token bus market leaders. Miller remembers:
"I said: ' Geez, you know, LAN really isn't launched. You know, I think we really ought to get our act together and have both sides of the business humming.' I mean, you go to do a public offering and that can be very diverting. So, I went on the road. It was similar -- the guy was telling me it was very similar to a public offering. I went on a road show. I went to Europe. To a whole bunch of cities."
Miller traveled most of June selling a $5 million private placement. Investors, however, were unwilling to commit until the NCC show proved successful.
At NCC, CDS provided the token bus controllers show participants needed to connect to the shared token bus LAN. It seemed to matter little that the connections were via an external box, not an internal bus-based controller, token bus worked. Concord Data Systems was a hit of the show. Miller easily remembers that the:
"show was euphoric. Unbelievable show. There was us, little us, you know, with only -- I mean they were all billion dollar plus companies and there was a big press conference: the first hoopla about MAP. It was a real splash. There were only seven vendors at the time: ourselves, DEC, IBM, Allen Bradley, Gould, Motorola and Hewlett-Packard. We were on top of the world!
One of our potential investors was at the press conference. He signed up for more after the press conference. We raised $7 million; two more than we wanted."
The $7 million financing closed in June 1984. CDS was valued at $34 million. Not bad considering only two years earlier the company's value was $8 million. But why shouldn't the value of their shares increase? They were working with GM, easily the largest potential token bus customer. DEC and Allen Bradley would soon be OEM's. There were only two clouds on the horizon: first, IBM was not focusing on the token bus opportunity the way they "should be," but they had been a 'bluebird,' a long shot, all along anyway. The second problem portended to be more serious. Phone calls to GE were not being returned. Even requests for meetings with GE personnel were being deferred. GE had to be converted to token bus and drop their Ethernet efforts. By delaying all they were doing was missing sales, and the opportunity to be a significant token bus player.
In the fall of 1984, CDS management felt like they were "on top of the world!." For the fiscal year ending September 1984, budgeted revenues of $12 million had been surpassed. Revenues totaled $15.5 million with marginal profitability.
CDS entered their 1985 fiscal year an organization filled with confidence. Now only to win the GM contract in the spring.
But problems soon arose. They couldn’t come out with the desired second generation V.22bis modem. Miller remembers:
"There weren't enough resources in modem engineering. It was sort of getting sucked up in LAN."
As important as the modem issues were, in the fall of 1984, CDS management had an even more significant problem. On October 10, a thunderclap echoed through the halls and offices of CDS, stunning management with the news that UB and GE had formed a joint venture, Industrial Networks Inc. (INI), to compete in the MAP-token bus market.
But again Miller rationalized:
"OK, they got a lot of money but, holy shit, we've been working on this thing forever!"