@Carnival-Boy said:
I don't think either are relevant or realistic in the real world. Yes, you can define the cost of providing certain HA solutions, but by definition HA doesn't prevent downtime, it only mitigates against it (hence the word "High" rather than, say, "Total"), so you'd still be asking "What kind of budget do we have to reduce our risk of downtime" and not "to prevent downtime".
And it is specific to "layers". For example, most people in the SMB talk about HA only in terms of the platform layer and ignore the apps, OSes, storage, power, cooling, access, network, etc. It's a myopic view. They mitigate almost no risk yet feel that they have moved from "so risky I'd never do it" to "so safe I never have to think about it" and yet, almost nothing has changed.
The entire drive for HA is often an emotional response. We have been taught to fear a server failing, so we over compensate and try to fix that one, moderately small risk without addressing anything else.