> Until the system goes down. Or a security issue stops the entire cloud system from working. Or a patch breaks everything.
I can't see there being a problem. IBM had this business model running really smoothly in the 1970s.
Yeah - but (generally) IBM then knew what they were doing. And a properly-run mainframe doesn't stop just because of a small hardware failure.. Especially when done in proper clusters with redundant hardware and sysops who know what they are doing.
 Software failures on the other hand. Well, lets just say that I once was a mainframe assembler programmer before I grew up and went into support.
 And haven't just come back from the pub after an (entirely) liquid lunch.