http://www-1.ibm.com/linux/news/semiconductor.shtml
So we set up three machines communicating to three pieces of equipment, using the same tool communication application on Red Hat 7.2 Linux and Windows NT/2000. We used the same setup for all six applications, and then we began a planned 90 day test. The plan was to run the applications and look at errors, performance degradation, maintenance requirements, and any other issue that might pop up.
The results were striking. The best run we had with the Windows machines was approximately six to seven days before we began to see noticeable performance degradation, and in about ten days we had a service interruption. If you multiply this failure rate times the large number of servers we wanted to deploy, you can see that with Windows NT/2000, it would have been a performance nightmare and could never have supported our overall fab goals.
The Linux servers, on the other hand, ran without incident for the full 90 days. There were no performance degradations, interrupts or required maintenance. In fact, just for fun, we kept the Linux experiment running to see what would happen. We finally shut it off after 5 months, and it was still running like a clock.
--
**************** *****************************
Kiggundu Mukasa # Computer Network Consultancy###
KYM-NET LTD. # Intranets & Internet Solutions#
House 73 # Data Communication Service ####
Plot 80 kanjokya Street
P.O. Box 173 Kampala, Uganda
Tel: +256 77 972255
+256 71 221141
Fax: +256 31 262122
*************************************************************************
|
