I think you're technically increasing your points of failures by splitting domain and hosting. You're not adding redundancies by splitting the services. Since

I think you're technically increasing your points of failures by splitting domain and hosting. You're not adding redundancies by splitting the services. Since these are mission critical services, if either fail, everything fails. Now by splitting up the services, you're increasing your risk of failure. Take this scenario as an example. Suppose a random host will fail 1 out of every 500 days. Well now you have two hosts that could fail on 2 random days out of 500 days instead of just one host that'll fail once for every 500 days. The above is not exact but i hope it gets the point across. I'm not saying dont split hosts, I do that myself. I'm just saying don't think you're limiting point of failure by dividing mission critical services apart. You need redundancy for that.