By default the API server only has 1 replica. If the POD fails, the single instance restarts by normal Kubernetes behavior, but there is downtime. If the user is upgrading, some downtime is expected in most cases as well.
Users have the option to configure the number of replicas using the app_replicas field in the scst-store-values.yaml file.
By default, the database has 1 replica, and restarts with some downtime if it were to fail.
Although the field db_replicas exists and is configurable by the user in the scst-store-values.yaml file, VMware discourages using it. The default internal database is not intended to be used in production. For production use AWS RDS. See instructions here.
For the default postgres database deployment (set by default or by setting deploy_internal_db to true), Velero may be used as the backup method. Read more about using Velero as back up here.
