<< Many of my enterprise-class customers have Windows NT/2K machines that haven't been rebooted in six months or more. For many of my customers, reboots are the result of OS updates and not stability issues. So if these same customers had chosen to defer OS updates, it's conceivable that their uptimes would be extened into a timeframe measured in years.
Of course, defering OS updates like that is not practical.
OS updates, however, can be scheudled. In this thread we haven't differentiated scheduled vs unscheduled reboots, but I think we'd all agree that the unscheduled reboots are of greater concern.
The number one cause of unscheduled reboots (in my experience) is blue screens. The number one cause of blue screens: antivirus software. (That, and HP 4000/8000 drivers on NT4 Terminal Server machines.) Interestingly, *all* antivirus vendors seem to be about equally guilty of crashing machines. The good news is that there are FAR fewer crashes from antivirus software than there were three years ago.
If you have performance problems when running apps, and you truly want to troubleshoot and resolve the issue instead of rebooting, fire up performance monitor and look for processes with extensive numbers of open handles, memory usage that keeps going up, etc. Although it's very rare that an app will truly bring down the OS, it's not so difficult for an app to consume so many resources that the OS is useless from a practical standpoint. If you find such an app and kill it, OS performance should resume to normal levels. >>
Very,very good points !!!!
I'm quite spoiled by the overall stability and robustness of WIN2K