These issues are in your control - not for the centralised service but your use of them. You can build appropriate redundancy for the components/providers in your stack and the budget you have.