We are currently working on benchmarking and optimising the stack. We are looking at collecting information about most of the metrics you have mentioned above. Specifically:
2) the stack is designed to scale horizontally, and our underlying storage layer is Cassandra. Although we haven't tested the max number of nodes yet, the expectation is that it will be large.
3) In terms of cloud we are going to support Oracle and Google cloud and it currently takes around 15-20 mins to deploy a running cluster