c
orporate networks. It might even have happened with a network computer; no one today can guarantee that network transmissions will succeed.
I didn't mess up my system running shareware, installing unauthorized software, or anything of the kind. I was running a configuration that's approved and stable 98 percent of the time -- and inscrutably, catastrophically troublesome the rest of the time. This isn't a function of the apps I use, the computer I use, or what I do with the system. It's a function of an architecture that has grown too complex with too little self-management.
Right now, my PC is sick. When you have a cold, a box of tissues (ZAK or the NetPC) is nice. But it's no cure, and a cure is what I want. So I respond to this summer's "reduced cost of ownership" drum rolls out of Microsoft and Intel with a sniffle and a yawn.
First, the NetPC. Can we get past the trendy moniker and just admit it's a cheap computer with no floppy (or one that can be disabled by software)
? Okay, it has a few cool things -- LAN wake-up, remote boot -- that will be available in most PCs in 1998. We've said many times during this whole cost-of-ownership controversy that one size does not, will not, and should not fit all, and this NetPC solution will fit only some needs. I'll even concede it's something of a step forward, but too small a step to fix the fundamental problems experienced by PC users and administrators daily.
The same can be said of Microsoft's ZAK. Much of it has been available as utilities from other vendors since Windows 3.1. Lacking those, you've long been able to turn off certain menu functions by editing .ini or registry files manually (scary thought). Thank you, Microsoft, for making a single tool to do that for multiple users. And the idea of policies and roles, ultimately tied to directory services, is spot on. But let's not mistake that for a solution to the underlying complexity of PC management.
Intel and Microsoft have created an environment that is wonderf
ully flexible and horribly cantankerous. Maybe that's the way it had to be, but it's too big a trade-off now. Instead of addressing the environment issues, Microsoft and Intel are essentially blaming the user for experiences like mine. Physician, heal thyself.
In 1992, Microsoft had a vision of an architecture that would address the ad hoc nature of its burgeoning environment and put it all on a solid, object-based footing. While the code name persists, Cairo has become a much more pragmatic and evolutionary (and still worthy) project. As far as ease of management, what remains of the original grand vision are some promising technology pieces that will begin to appear in Memphis and reach full fruition with NT 5.0: limited self-healing and self-updating capabilities for applications, the ability to store machine state on a server, intelligent local caching, and, finally, forbidding app vendors from installing OS components willy nilly. Good ideas all, and in many ways more generally applicable than the
pure network computer. In any event, those technologies sure sound a lot more like what we need than ZAK does. Serious inroads into Windows admin costs will depend on their success.
Gee, I've gotten to the end of this piece without another crash. Maybe it was just that quirky keyboard connector -- I'll never know for sure. Certainly thousands of engineers at Microsoft and Intel can do something to find out besides making a computer dumber.
Mark Schlack, Editor in Chief,
mschlack@bix.com
The NetPC Blues
