I think at least EWC gateway should be available at multiple independent data centres (i.e. ECMWF and EUMETSAT side of the EWC) with the same content to ensure sufficient availability.
Also, the gateway should work without single point of failure or at least via cache effectively APIKEYs / tokens / other information coming from single point to mitigate possible effects of one module being in troubles.
Design the architecture to work even if one site is down and test its functionality.
EWC already contains some building blocks for the solution. Especially:
1) Rancher Kubernetes manager which can manage Kubernetes clusters in multiple clouds (including AWS)
2) Route53 DNS
If these are needed, I'm happy to setup a meeting with the developers to setup the environment.
(EWC is also developing managed Kubernetes Engine which should cover these cases even easier but the timeline is probably is probably not in sync with the FEMDI project).
I think at least EWC gateway should be available at multiple independent data centres (i.e. ECMWF and EUMETSAT side of the EWC) with the same content to ensure sufficient availability.
Also, the gateway should work without single point of failure or at least via cache effectively APIKEYs / tokens / other information coming from single point to mitigate possible effects of one module being in troubles.
Design the architecture to work even if one site is down and test its functionality.
EWC already contains some building blocks for the solution. Especially: 1) Rancher Kubernetes manager which can manage Kubernetes clusters in multiple clouds (including AWS) 2) Route53 DNS
If these are needed, I'm happy to setup a meeting with the developers to setup the environment.
(EWC is also developing managed Kubernetes Engine which should cover these cases even easier but the timeline is probably is probably not in sync with the FEMDI project).