"File server was the most difficult thing in the world to communicate," recalled Judy Clark. "And we never could get the point across. We would say, `Yes, we're a file server and this is why we're a file server, and they're not -- they are a disk server.' And so we'd get this all straightened out with [an] editor, and then [the editor] would go talk to somebody else, another company.... And they would say, 'Yes, we have a file server [too].' And so then the editor would [write], 'Yes, these two products are file servers.' And they were completely different kinds of products."
Then one day in September 1984, the heavens opened and the revelation was handed down to the mortals below.
"The big break for Novell was when IBM endorsed Microsoft's PC LAN v1.0, an early [file-server-based] LAN product," said Neibauer. "Until that point, Novell was fighting an incessant battle with 20 other small manufacturers about whether file server or disk server technology was a better technology.
"When IBM did that, all of a sudden the 20 other companies who had been trying to sell disk servers withered up and blew away. Almost all of those -- those that didn't go bankrupt -- within a couple of months came and started buying Novell stuff and selling it, because IBM had now spoken. The giant had said: `We will have file servers. We will not have disk servers.'"
So Novell found itself on the side of the angels.
A consequence of the file server model developed by Superset was "hardware independence": the ability of NetWare to function independently of network hardware. In a general way, the idea of hardware independence had been around for a long time. Jack Davis had recognized the value of being able to link different operating environments on the same network; he dreamed of linking Unix computers with CP/M computers. In 1982, Superset succeeded in connecting DOS and CP/M on a working LAN. Although the potential of hardware independence was appealing from the beginning, only in late 1983/early 1984 did Novell begin to explore and exploit this concept commercially. As it turned out, the company's marketing strategy based on hardware independence was the key that helped Novell dominate the market in just three years.
The Novell that Ray bought into in 1983 was a systems manufacturing company -- a company that produced primarily hardware with software bundled in. Nearly all of the income in 1983 was from sales of the NetWare/S-Net system and related hardware components.
In the 1960s and 70s, IBM and other computer companies had established proprietary computer systems designed to lock in customers. The computer manufacturer tried to be the single vendor for all the customer's equipment, supply, and service needs. As the customer's needs changed, the single vendor would lead him down well-established migration paths to bigger and better systems. Compatibility with other systems was discouraged.
This single vendor model was the goal of virtually all of the early LAN companies, who sought to provide a total solution to the customer's needs. In 1983, Novell was just one of many companies offering proprietary LAN systems. Novell offered a "turnkey" system that included a file server computer, network interface cards, cabling and connections, hard disk subsystems, and the NetWare Operating System. (Novell did not sell PC workstations.) As a manufacturer of a proprietary LAN, Novell faced an uphill battle against Corvus, whose Omninet already dominated the LAN market.
Novell at the beginning of 1983 wasn't a strong company, and it was just one of many companies offering networking products for personal computers. Whatever slim chance Novell had of carving a dominating niche in the hardware end of networking of personal computers sunk forever in 1982's stormy sea of red ink and crisis.
Thus in the course of doing business that first year, Ray and his team soon realized that by limiting the sale of NetWare to Novell LANs, they were limiting the growth of the company. If NetWare was ever to become an industry standard, it would need the kind of mass distribution that could only be achieved through licensing: to gain the "critical mass" of installations, NetWare would have to be bundled with other LAN systems besides Novell's S-Net system. The choice was between "rifle shot" marketing, where NetWare would be carried to the market in a single slug (S-Net) or "shotgun" marketing, where NetWare would be carried to market in a shower of vehicles. To make a massive impact in a short time, Novell decided to use the shotgun.
Craig Burton recalls how this vision came to him.
"As soon as Ray joined the company, I was going to move from Marketing into Sales. Ray convinced me that I should stay and be director of marketing.
"It was sometime in early 1983, I think, that the [PC]XT was announced.
"Drew and I both felt that the competitors were Corvus, Orchid, and a new company called 3Com. PC-Net (Orchid Technologies) was probably growing faster than all of them at the time. Drew and I both thought the XT would be a terrific server and that we should start being as hardware-independent as we could -- not drop the hardware yet -- just become hardware-independent. Ray was supportive of it and allowed us to buy the equipment to do that.
"One of the questions we were struggling with at the time was: What network should this hardware-independent version of NetWare run on? We hadn't come up with being network-hardware independent, we'd only come up with targeting NetWare [server running] on the XT.
"The question kept coming up that we needed to pick a medium to run on. Should we build a product that's compatible with the Novell 68000 box, or should we find someone else's network adapter? The choices we were struggling with were: Should we do Corvus? Should we do 3Com Ethernet? Should we do Orchid PC-Net? Should we do Arcnet? Those were the four big choices at the time. Or should we do our own?
"I sat down and listed all the companies that were these company's vendors -- Orchid had the biggest list. I think 3Com was next. The other thing was that Ray was a big investor in this company in southern California called Gateway Communications and they had yet another network called G-Net with no network operating system on it. Because Gateway was willing to cooperate with Novell at the time, that was the first we started working on.
"But I can remember very clearly when I finally came up with the answer in my head about which network we should support. And the answer was: All of them!
"I was driving to the airport and I pulled over and drew a box that had all these different network cards in it and all the companies that were supporting the networks attached to them. So I had an Omninet, an Ethernet, and an Arcnet line. At G-Net there was no one but us at that line.
I took it back to Drew and said, 'I want this box. I want an XT to have all these cards in it and have them look the same.'
"He said, 'I can't do it.' (Of course, several years later, he did.)
"I said, 'Well, okay, we'll pretend like we did.'
"We went out and started creating -- I think that the other major event that caused a significant turn in the mentality of the company. Coming up with the notion that all of the hardware, including the server and the network adapter, was irrelevant to us. That we would do all of them. We had a lot of resistance to that at the company."
Making a networkable computer in 1983The key to an easily networkable computer in 1983 was whether or not it had a well-documented expansion bus. An expansion bus is a socket or series of sockets on the main computer circuit board (the mother board) where other circuit boards (daughter boards) can be added to expand the computers capabilities. This is the idea of modular design applied to personal computers. The first generation personal computers were all built with an expansion bus on the mother board. The uses to which these early computers would be put were so varied that the designers needed to make them easy to customize. The easy way to customize was to add different boards, so they put in a place where many different kinds of boards could be added -- the expansion bus. An expansion bus can be a single sockets -- as it is on the MAC SE. The most prolific were the early IMSAI's and Cromemcos which had 19 and 21 slots respectively. These were so modular, in fact, that the mother board consisted of nothing but expansion slots -- all the RAM, all the I/O and even the CPU came on daughter cards. But modularity adds expense and size to products. By 1982 the CP/M based computers that were coming out were integrated -- they had everything on the motherboard and no expansion bus. These new machines cost less and looked more stylish, but they couldn't accept a networking card. The IBM PC followed the Apple II tradition and stayed modular. This made the PC exciting to people like Drew Major who wanted to do more with a computer than just connect it to a printer and a modem. |
Novell's S-Net was a product of its times. Engineers and marketers had discussed other alternatives but S-Net seemed best suited.
In 1981/82 there were no networking standards. The magazine articles of the day were still discussing the theoretical virtues of baseband versus broadband and various networking topologies such as linear bus and star because there were few real networks to discuss.
Novell considered Ethernet, but in those days it was much too expensive -- over $1000 per connection -- and "thin" Ethernet hadn't been developed so the only way to connect was with bulky converter boxes connecting equally bulky "thick" coaxial cable.
Arcnet was also explored and rejected because of it's cost.
The end result of this elimination process was S-Net, an in-house design. S-Net used a star topology with all the workstations connecting directly to the file server. It used an inexpensive to design and build RS-422 electrical protocol that was carried over dual twisted pair cable (4 wire phone cable).
S-Net was attractive for it's simplicity. All the cables had only two devices on them. This meant there was no need for a sophisticated communications protocol that had to provide for addressing different stations on the wire. The file server knew which workstation it was talking to simply by knowing which communications port the signal came from.
S-Net was yet another quick and inexpensive choice on the headlong rush to the a LAN to the market. It was a good choice at the time, a time when LANs where self-contained systems of half a dozen workstations and devoted to sharing hard disks and printers and little more. The S-Net concept would serve less well as the LAN concept expanded to embrace larger networks and connectivity to many other kinds of data processing systems.
As the LAN concept evolved to include connecting larger networks and interconnecting networks, other needs appeared, and those needs would be better served by other architectures. Novell had a hard choice to face: How was it going to expand it's market beyond S-Net?
The decision to market the software component apart from Novell's hardware required a certain leap of faith; it was a bit like stepping off into the void. For one thing, all the company's money came from sales of the S-Net system, which was mostly hardware, and other hardware items like network interface cards (NICs). The only thing really unique about S-Net was the software operating system. If other companies could buy NetWare, wouldn't that kill off Novell's hardware sales -- and therefore Novell? Wouldn't Novell be shooting itself in the foot by OEMing NetWare?
One of the most passionate and articulate advocates of unbundling the NetWare operating system from the S-Net product was Craig Burton. In working with early PCs, he had seen how PC manufacturers could use third parties to leverage sales of their products. One example was the Apple II, and the other was the IBM PC. The Apple II did very well in the pre-IBM PC world because it was a very open system. The design specs for the bus and monitor were well-known, so third parties could come in and build boards to their heart's content. In that period, you could do more things with an Apple II than you could with a CP/M machine, even though there might be more power in the CP/M machine. Variety was definitely the Apple forte, and that's why it sold so well.
When Apple came out with the Apple III, they closed up the bus. And when they shipped the Lisa, they kept it closed. And when they came out with the Macintosh, it was still closed. The people swelling the ranks of the growing Apple Computer Corp. fell in love with their graphic operating system, and lost track of the importance of an open hardware system.
But Phil Estridge, who headed up IBM's PC business, did not. He talked to Computerland, and he talked to a lot of other people as IBM rushed its entry into the personal computer marketplace, and one of the things that became very clear to him was that the open bus was an important selling tool. Technologically, the IBM PC is the successor to the Apple II, not the Macintosh, because both were open systems. Third parties who had worked and developed on Apple II looked for the next machine to go on to, and most often they went to the IBM PC rather than the Apple III or the Lisa, because neither of those were open systems.
Burton brought this vision to Novell: When you're playing in the PC world, the ability to let other people play with you is of paramount importance. This was also Superset's vision, and Ray picked up on it quickly.
Novell's weak position in 1983 reinforced the concept -- leveraging sales of your product off the backs of your competitors was the only practical way for a small company to attain market acceptance in a short period of time. A year in the high tech industry is like a decade in other industries, and if a company fails to make an impact on its market quickly, its hour may pass, dooming it to second-rate status or extinction as its competitors gobble up market share. Burton called the game of strategic alliances "corporate jujitsu." The successful player would turn the force of its opponents to its own advantage. In the years to come, Noorda and Burton would prove themselves masters of corporate jujitsu.
In 1983, Novell evolved from thinking of itself as a LAN systems company (hardware/software) to thinking of itself as a LAN operating system (software) company. It would continue to manufacture and sell hardware as a way of promoting the software -- and as a way of surviving, because software sales were nonexistent in 1983 -- but the strategy for growth would focus on the software that ran the network hardware.
Neibauer described the reasoning that led Ray to take Novell down the software track:
"The hardware independence business came about because there was [in 1982-84] an incredible proliferation of small companies who were selling lots and lots of hardware solutions, and Novell was just one in a crowd.
"The thing that made us a little different was that we owned our own software. And it ran better than most of the other software offerings that people were making right then.
"Corvus, NestStar, Televideo: There was a huge number of them, and a big number of these people ended up becoming Novell OEMs. Most of those people started as Novell competitors. There were 30 or 40 different companies who were all trying to push PC networking solutions. Most of them had proprietary network hardware, had proprietary board plug-ins, and proprietary software.
"And that's why we went hardware independent. We looked at it and said, 'You know, our hardware is not really that cost-effective. Let's face it guys: We've got some problems. But if we could leverage off our software, then every time one of our competitors makes a sale, we get a little money. That will keep us alive.'"
By the end of 1983, Novell was trying to market the NetWare Operating System independently of its S-Net LAN. This effort entailed:
Craig Burton was the principal evangelist*** for NetWare, with help from Ray and Superset. The effort really got underway in 1984.
[* "OEM" stands for Original Equipment Manufacturer. An OEM company buys components from other companies to include as part of its manufactured products. For example, in 1983, Novell was an OEM company that bought Motorola chips as components to be used in the NetWare file server. "OEM" is also used as a verb.]
[** "Porting" means moving software from one computer architecture to another. A small part of that process is making a communications "port" attaching the two systems so that files can be transferred.]
[*** "Evangelist" is a term applied by Apple Computer to its marketing and developer relations people. Their job was not merely to sell and persuade but to spread the gospel.]
Support for this concept was far from universal within Novell. Jim Bills pointed out as late as 1986 that, "Novell makes more money selling its hardware than it does its software. If we're going to keep growing we can't abandon the hardware until the software can carry the load." Harry Armstrong concurred, "It's a good concept, but it can't happen too quickly." And there were a lot of Engineers that felt it was a good idea as long as they could keep improving and releasing new versions of S-Net and the 68B file server.
If Novell was a bit unsure about unbundling NetWare software from its hardware, its competitors were even more skeptical. Judy Clark remembered how resistant the companies were to OEMing NetWare:
"And by that time we had figured out that the only way to really be successful was to be in every distribution channel and to be on everybody's hardware product. And so we had put together this campaign that ... internally it was called Project Piranha because we were just going to eat up everything. And it was to port the software over to everybody else's hardware....
?I had gone on trips with the two of them [Noorda and Burton] to try to explain to these people what the product was: that it was a file server, not a disk server, and that it could work on everybody's hardware. . . . If they would just dedicate some manpower to it, then they could do it. Then they could have NetWare on their products too!
"It was just like talking to a wall. Everybody was so territorial and protecting everything they had already come up with. They didn't want to cooperate. It was a really immature industry, and no one could see how big it could really be.
"Other companies didn't see how they would benefit by helping to proliferate Novell's LAN operating system. In classic computer industry style, they were still trying to deliver a total solution. And it was by no means clear to anyone at that point that NetWare was a superior technology or one that would win wide market acceptance."
The industry's lack of interest in Novell did not deter Burton; he proselytized relentlessly.
In April 1984, after weeks of negotiations, Novell lined up its first OEM customer, based in Atlanta, Ga. Shortly after, a second agreement was made with Gateway Communications, the company Ray owned a piece of. Others followed. "We went from one and then two, you know, it was really slowly on the third one, and then we had six," remembered Judy Clark. "It was thrilling. You know, there are certain moments in the history of Novell where you're struggling, struggling, struggling, and all of a sudden you have six OEMs!... That was a big, big break...."
The six OEM customers were Sperry Corporation; Quadram Corporation; Televideo Systems, Inc.; Texas Instruments Incorporated; North Star Computers, Inc.; and Santa Clara Systems, Inc. The breakthrough had come partly through a combination of diligent work and the felicitous support of IBM for the file server technology of the PC LAN v1.0. As noted by Neibauer, after the IBM announcement, NetWare looked a lot better to Novell's larger competitors who had placed losing bets on the doomed disk server operating systems.
As Noorda and Burton worked to win industry support of NetWare, Superset labored to make hardware independence a reality. Achieving it created an enormous drain on Superset's development time and energy. The job of creating shells in NetWare to support different kinds of PCs, different PC operating systems, and different types of LANs occupied Superset from 1982 until well into 1984. In 1984, with Project Piranha, NetWare had to be ported to all the major network hardware. And while they plowed through the hard ground of compatibility problems, the Superset guys were also working on a new version of NetWare. The new version, called Advanced NetWare, would allow users to add multiple servers to their LANs and to connect different LANs via "bridges."
* [Bridges, routers, and gateways are three different means of connecting networks, and the terminology was not standardized in this era, so Novell had different meanings for these terms than, say, Cisco does today.]