Hi all, I'm trying to find the most efficient way to find a node in the cluster (from the client side) that may hold a copy of rows I'm querying for. The scenario is quite simple: I have a hadoop job which reads an index and then has several thousands of keys, now I want to find a way to efficiently retrieve the according rows from the cluster, for this I have to find the node that is responsible for the key. Am I right in the assumption that I can just calculate the MD5 hash (in random partitioner) of the keys, order them according to their hash, then order the TokenRanges by their endToken and then do a merge sort?
Is there a faster way to do this? Regards, Chris
