Just as a group of young university students on a recent highbrow quiz show had never heard of Frank Zappa and the Mothers of Invention, so one day we must resign ourselves to the fact that, one day, no-one will remember NetWare in the context of Utah-based Novell. Or will they?
As Novell marches towards the release of its first open source community-oriented project, OpenSuse, the once mighty Novell whose coffers were distended by cash from the sales of a red-boxed network operating system (NOS) that ran most servers in the world, is a shrunken husk of its former self.
Today, the NOS market niche that Novell almost monopolistically filled has disappeared, papered over with blurred distinctions that make the network a much more complex place than it was even ten years. And that's despite there being only one protocol on most of today's networks; in those days, you'd be sure to find IPX and one or two proprietary IBM protocols, as well as IP.
Back then, all you needed -- or so it seemed -- was a Certified NetWare Engineer (CNE) qualification on which to build a lucrative career. Hardly surprising given the complexity of the NOS.
For instance, with NetWare 2 -- effectively a rebadged version of NetWare 286 which followed the launch of the Intel 80286 chip -- you had to recompile the kernel to make a configuration change. By the time it got to version 3 -- aka NetWare 386 -- the difficulty of interfacing it with hardware, and loading and running applications was to a degree ameliorated by those pesky plug-ins known as NLMs -- NetWare Loadable Modules.
NLMs consisted of code that didn't just load into a protected area. Instead, they operated at kernel level, with all the potential for instability that entails: one memory bug and the whole system keeled over.
But then, those were the kinds of risks required when PC hardware was barely up to powering enterprise-level servers. In today's world of pre-emptive multi-tasking, where the OS manages processor time, Overlord style, it's easy to forget that one of the reasons NetWare's performance was head and shoulders above that of its rivals was that it depended on co-operative multi-tasking -- ISVs' code had to allow other processes to have a slice of CPU from time to time. Hardly surprising then, that the result was lock-ups and freezes as processes vied for the attention of the processor, as each vendor's code co-operated to a lesser or greater extent. What it did mean was that the NOS added little overhead and, as a result, the system flew in comparison to chunkier systems, Windows included.
But once the system was up and running, the stories about the NetWare's stability became the stuff of legend. Some talked of servers that stayed up for years without human intervention -- as long as no-one tried to 'fix' them.
Internet and hubris
But all that counted for little in the face of two key changes. First was the hardware's burgeoning power, making software efficiency increasingly less of an issue. Publications such as PC Magazine which used labs equipment to measure application performance gave up doing so during the 1990s as users and buyers gave up caring about this metric. Software was no longer the bottleneck.
The other big killer for Novell and NetWare -- for the company was still largely a one-product organisation -- was the power of the Microsoft marketing machine.
Few argued that Novell had superior technology at the OS level with NetWare, and the directory level with NetWare Directory Services (NDS). However, as the importance of IT grew within enterprises, the locus of purchasing power moved up with it. Instead of the geeks being specifiers, board members such as the CIO signed the cheques. At this level, they were a lot more susceptible to Microsoft's messages.
Novell helped though. It took its eye off the ball in the mid-1990s, big-time. At a time when prescient companies were heading for the Web, Novell bought Unix kernels and applications from WordPerfect in a hubristic bid to take Microsoft out. It failed and, within two years, Novell had sold its acquisitions at a loss. Meanwhile, Microsoft had impressively turned the company round on a sixpence and was becoming Web-focused in a hurry.
Novell's attempt to move into the services market proved to be a bit too late and maybe not enough. NetWare became less relevant as Windows acquired Web-based features and, with the launch of Active Directory, Microsoft appeared to have Novell in a head-lock.
That hasn't stopped Novell continuing to acquire and develop new products for the emerging Web services market and today, it has moved its focus onto Linux via the acquisition of SuSE and this year's release of Open Enterprise Server. It offers services using one of either the NetWare or SuSE Linux Enterprise Server 9 kernels.
Novell insists that NetWare will remain on sale and supported as long as customers want it. And while the company's installed base is predicted by IDC to drop by almost half to under 1.5 million servers by 2008, it would be a fool who wrote off Novell. Only NetWare's long-term future seems in doubt, given its lack of USP for new buyers.
And Novell itself appears to be road to recovery. It's even been turning in an occasional profit recently -- and that's no bad start.