There is no 'self-healing' system. Nothing can detect a 'real' failure as opposed to a counterfeit failure without heuristics that depend on an interpretation

There is no 'self-healing' system. Nothing can detect a 'real' failure as opposed to a counterfeit failure without heuristics that depend on an interpretation that is inherently fallible. Ops was about discipline and method. Restrict the problem set to known values. The one thing that I see these days is that developers don't have a fucking clue about ops but they do see that the basic task responsibility is toilsome, counter-creative and that 90% of it can be automated by reliance on coded procedures in write once argot that hopefully is editable and doesn't give up the family jewels when posted in pub github.