Cloud services and resilience

After a Facebook discussion this RNZ article on Microsoft wanting to push AI systems in New Zealand schools, I got into a conversation with a local hydrologist (also Leon's mum) about cloud services and resilience, especially in light of the devastation wrought by Cyclone Gabrielle recently.

Marianne writes:

I'm genuinely curious as to what the fallback options are when your primary cloud services go down in a situation like Gabrielle and all your data and software are in the cloud and reliant on the internet working.

Monitoring sites are usually independently powered and have battery banks to keep them going. Critical sites are doubled-up. We used to run the comms networks with our own repeaters with backup generators, backup radio comms in case the copper or cell towers went down, backup generators at the EM centre, and a local PC with the necessary data and software installed so if we had to, we could run the entire system for several days from pretty much anywhere.

I keep hearing about cloud services dropping the ball in these situations. So as I said I am really curious and interested to know how cloud service providers can ensure this sort of resilience.

The very short answer: they can't. Not by themselves.

( The relatively long answer )