The battle for the desktop was not won by Windows in any incarnation, but rather by Unix. While this statement may sound bold to many, it is not surprising to find that it emanates from the mouth of Doug Michels, executive vice- president of Unix-vendor, The Santa Cruz Operation.
Speaking at the MUA 96 conference in Leura last month, Michels said the current shift towards network computing promotes all the ideals of open systems. "The Internet way of computing predicates a desktop that is standards-based, not a desktop that is de facto open systems-based.
"That desktop can be any device at all, and they are going to meet the NC reference profile, and they are all going to work with your server applications. And those server applications are going to run on Unix." Michels says that many of the standards the Internet is based on, such as TCP/IP and e-mail, were developed using Unix as a base operating system.
Selling the story
He said there are two arguments that will sway corporate users in favour of NCs. Block-mode and VT100 terminal users will be swayed by the ability to add greater functionality to their applications through pointing devices and improved graphics. "An Internet-based style of computing leads us to a model where we can get the end-user productivity that comes from graphically-oriented systems and higher-end applications, more than we could ever on dumb terminals. Unlike the PC it will give us the ability to centrally manage and run these things at a reasonable cost."
The second argument is reduced management costs, which will win over fat client-users. Michels says a Gartner Group analysis of total cost of a Windows-based PC ownership in a major US corporation was $US9,000 per year per seat when considering full infrastructure and support costs. He says the estimated costs for an NC in the same environment are decreased by two-thirds, at $US3,000 per year.
Michels says that in light of such equations people should also stop thinking about the hardware costs, which he admits in many cases may be much the same for both thin and fat client computers. "Even if the cost of acquisition has zero saving, it won't affect the $9000 very much. The real issue is cost of ownership. Cost of ownership is driven by the fact that an NC is a zero admin client." With no persistent local storage, Michels says there is less chance for something to go wrong on a thin client; at the same time eliminating the cost of local software upgrades.
He added that as the NC does not need to defer to current design standards there is greater room for innovation.
"Because the NC is designed in terms of standards at an abstract level, it allows people to innovate in the hardware. You say we don't have to use an Intel chip, we don't have to use Windows, we don't have to put in the same connectors, we have a lot more freedom. Good designers can find ways to save money if they can freeze the definition and work underneath it."
Michels claims all of these factors combine for the triumph of open systems. "We thought the way to get open systems to win was to put Unix on the desktop, but that isn't the right answer - you put standards on the desktop, you put Unix on the server.
"The client/server model has proven itself for office automation, productivity applications - everything but business critical applications. It has not worked, and the cost of doing it has been much higher than anybody ever thought."