So far so good, I don't expect the performance to be fantastic, it just should work. To my big surprise, the stats in the backup job say that "proxy" is the bottleneck - how could that be? It should be clearly the network. Is it just a misinterpretation or is it true??
Please open a support case and ask our engineers to analyze logs and performance statistics from all nodes of the backup infrastructure. I cannot make any conclusion based on the diagram above only but the word "proxy" here most likely means the source server load (on which VAW is running) as the main compute operations (f.g. compression) are carried out on this server.