Closed jiayihu closed 3 years ago
I realized there's the need to have a local database as a service from the service broker in order to save the rainfall time series. For the time being, I can write a key-value database myself and provide it as a service to the cluster #67 .
Data still needs to be sent to the cloud to be later used for ML training. Using a cloud database requires a specific client library, e.g. google-datastore-rs or google-cloud-rs.
That would make the application not independent from the service provider. Ideally one would wish to use any storage available as a service.
Sending data to Datastore, provisioned as service and with binding credentials as k8s secrets, is working. google-cloud-rs
is not maintained anymore and there isn't a better alternative, but at least it's still working after updating the dependencies to tokio 1.x https://github.com/google-apis-rs/google-cloud-rs/pull/37
Epic #61