Interface DataPartitioner


  • public interface DataPartitioner
    Creates data partitions to allow parallel processing. Each data partition is defined as a list of segments, thus no segments can be in more than one partition and all the partitions together must cover all segments.
    • Method Detail

      • split

        java.util.Collection<java.util.Set<java.lang.Integer>> split​(CacheTopologyInfo sourceClusterTopology,
                                                                     int partitionsPerServer)
        Parameters:
        sourceClusterTopology - the CacheTopologyInfo of a cluster where data is located.
        partitionsPerServer - how many partitions to produce per each server. In some circumstances, it may not be possible to honor this param, for e.g., when there aren't enough segments compared to the number of servers.
        Returns:
        a collection of data partitions