It would be great to choose storage per node.
It would be nice to specify an specific machine-id for foundationdb per node because of proximity, as well as an specific server/region for S3 per node.
The goal here is redundancy. If a stalwart node can’t reach an specific storage node, it would be nice if it became unresponsive so that other nodes would lower MX priority would take it. This would make the stalwart node more resilient and efficient since it would use the closed node available for s3 and foundation, reducing latency as well.