Author here - it's not that we're detecting failure better than they are (though certainly, we might be able to do it as fast as anyone else) - it's what you do afterwards that matters.
Being able to fail over to another database instance backed by a different volume in a different zone allows for a minimization of impact. This is well inline with AWS best practices, it's just arduous to do quickly and at-scale.
Being able to fail over to another database instance backed by a different volume in a different zone allows for a minimization of impact. This is well inline with AWS best practices, it's just arduous to do quickly and at-scale.