A lot of these are just "garden variety" (distributed) systems problems. Dealing with systems with differing latency distributions, recovering from failure, acc

A lot of these are just "garden variety" (distributed) systems problems. Dealing with systems with differing latency distributions, recovering from failure, acceptable tradeoffs between speed and accuracy, etc