We have decided to use a combination of our own infrastructure and cloud services (AWS, DigitalOcean). Even-though we could have run the entire Chainlink infrastructure on AWS alone in our opinion it makes sense to use atleast one of the other cloud providers (Google Cloud Platform, Microsoft Azure, Digital Ocean, Linode) or your own infrastructure in case AWS will have multi region or global failure.
Each set of nodes is geo-redundant by being hosted in different regions by different providers. Our own servers are hosted by a hosting provider certified in accordance with DIN ISO/IEC 27001. It also offers state of the art DDoS protection and multiple 100 GBit/s peering points on the backbone with guaranteed bandwidth of 1GBit/s.
We have a monitoring in place along with a person on call duty to make sure we can respond to unforeseen events as soon as possible. We are planning to integrate Grafana and Prometheus to our infrastructure in the upcoming weeks as well.
Current picture is taken from Grafana demo dashboard.