To: Justin Banks who wrote (16925 ) 2/1/1998 11:14:00 AM From: Bill Jackson Respond to
Justin;Do you have any facts to back this up, or is it just a gut feeling? NT scaleability (SMP, not clustering), is quite limited, at least partially due to it's close similarity to Mach. I believe that NT's success is at least partially due to the hoodwinking of MIS managers, ala MSFT's 'scaleability day' last year. If you tell them that NT can save them $x/year TCO, they're going to believe it. What they don't realize until later is NT's 'cost creep'. Sure, it's a cheap system, but to approach the same level of utility that you get with your typical Unix system you've got to spend a couple of thousand on third party stuff and other add-ons. While this is likely to get better with NT5 (if/when it ships), it still won't approach the usability, maintainability, and scaleability of Unix. Administrating an NT system is a nightmare, and consumes much more resources than Unix. You get back to me when NT gets TPC numbers in the 40k to 50k range, with a single system image (no clusters, they're a nightmare), and we'll talk. Sure, it will happen, but when, and how much more will Unix systems improve by then? What I hear you saying is that NT is soon to be a 'one size fits all' solution. This is madness, and will only serve to further impede the progress of technical computing.
