A while back, M$ trumpeted the news that the London Stock Exchange was using Lose 2003. Here is the result. They crashed and were off the air for over 6 hours. Makes my little problems with a memory leak in a print server by M$ seem tame. I can predict my crashes and reboot… The LSE has a pool of IT talent and they took hours to get things running. Rebooting 100 servers does take time…

The fact is that if you rely on computers (Who does not these days?) you should not rely on M$ to provide the software. It is too complex, bloated and impossible to debug in a finite length of time.

I am system administrator of a tiny system with 7 servers. In four weeks we have had 2 system-wide failures to share files, three failures to print and several wierd incidents involving “profiles”. Instead of having the whole system run smoothly on one or two servers, it has evolved into a monster which no one can fix. I could replace all its functions in a weekend but users are afraid of change. Better the devil you know than a new OS… Instead of typing passwords a few times a day, I am typing ctrl-alt-delete and passwords all day long and all I can do is watch it fail. My latest estimate is 72 hours to failure. Should I reboot tonight or gather more data?