k but a) back pressure as per my original post with Elixir (or Apache Storm, or Flink, amongst many others), and b) if you need latency-aware cross-timezone you

k but a) back pressure as per my original post with Elixir (or Apache Storm, or Flink, amongst many others), and b) if you need latency-aware cross-timezone you should probably be letting another piece of distributed infrastructure, explicitly designed for this, handle the issue, for example local/cluster/rack/data-center aware Cassandra. There is never any reason to re-invent something which big, multi-national corpos have worried about for you for a long time already. It doesn't strike me that anything in the OP post has not already been solved for 99% of us, using the right, freely available and well-documented tools. Distributed computing is hardly a new technology.