Presto router provides multiple scheduling algorithms for load balancing across multiple clusters.
Randomly selecting a cluster from a list of candidates.
Selecting clusters from a list of candidates in turn. Note that as the algorithm keeps the state of the selected index, it can only be used when the candidates are always consistent.
Selecting a clusters by hashing the username. This ensures queries from the same user will always be routed to the same cluster.
Randomly selecting a cluster from a list of candidates with pre-defined weights. Clusters with higher weights have higher opportunity to be selected.