Paul, Re: "If their existing systems were "SO STABLE", why did they have to think about
UPTIME and SYSTEM PROBLEMS EVERY MORNING ?"
Not sure exactly what that means, but MTBFs of S390 systems are now quoted in terms of decades. 100,000 hours, which is more than ten years, was passed by IBM CMOS machines a while ago. They're measuring upwards of 20 years now. Of course, that's calculated, based on a few thousand machines' statistics, individually measured in terms of MTBF over periods like 13 or 26 weeks. The servers we talk about here are really in their infancy WRT reliability. I like the way Intel and their alliance partners are addressing RAS, including system management over the network, hardware redundancy, failover, etc., but it's just a start.
Tony |
