The coordinator can optimize latency for a SELECT by asking data from the
lowest-latency replica using DynamicSnitch. It's not really load balancing
per se but it's the closest idea.

Reply via email to