It seems amazon had a cloud outage of up to 72 hours.

It was something to do with their storage servers, something failed in one data centre, it started to re-replicate from other data centres.

The re-replication overloaded other storage servers, and the network between them.

The outage then began to effect more data centres, and more customers.

The worst hit customers have permanently lost some or all data.

I guess its the adage ‘don’t put all your eggs in one basket’. Some sort of offsite backup is obviously the answer.