On 9/10/2013 7:43 PM, John Gilmore wrote:
He then turns to extolling the reliability and availability of blade servers, and here we disagree. Whatever their other deficiencies may be, mainframes are much more available and reliable. Interestingly, this is the case not because mainframe "hardware failures" do not occur but because they are recovered from when they do.
Did I mention blades? I don't think so considering we run rack servers, and very reliable they are too. You missed my point as usual. Hardware failures are not as common as they used to be but human errors are. I can think of some very major computer system failures in the last 5 years that were on mainframes. The Air New Zealand generator test disaster, the National Australia Bank batch restart fiasco, the Royal Bank of Scotland CA7 disaster. All were due to human error. Microsoft Azure had a 12 hour outage because they let an SSL certificate expire. It doesn't matter how reliable your hardware is, if humans are involved you have a point of failure. x86 lost it's unreliable reputation years ago when IBM started putting enterprise class chipkill RAM modules in it's X servers. SAP are quite happy to run enterprise SAP on Xeon. The introduction of machine check architecture on Xeon closed the gap even more. You probably need to get out a bit more.
---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN
