Practical Real-time Data Processing and Analytics

Combiner aggregator

The following are the characteristics of the combiner aggregator:

• It returns a single field as output
• Partial aggregations before transferring tuples over the network
• It executes an init function for all tuples
• It executes an combine function for all tuples until a single value is left
• Interface available in package:

public interface CombinerAggregator<T> extends Serializable {  T init(TridentTuple tuple);  T combine(T val1, T val2);  T zero();}

Example:

public class Count implements CombinerAggregator<Long> {  public Long init(TridentTuple tuple) {    return 1L;  }  public Long combine(Long val1, Long val2) {    return val1 + val2;  }  public Long zero() {    return 0L;  }}