OK, thanks for clarifying. Still, the point being illustrated was that a drop-in solution is rarely feasible... ie. the common HA cluster feature of redundant link-layer communications paths does add significant protection against availability loss such as the situation you describe. It's not just a software thing.