Apache Flink：keyby and window operator

Question

I want to know some mechanisms related to keyedstream. The code is as follows：

DataStream<Tuple2<String, Integer>> counts =
            // split up the lines in pairs (2-tuples) containing: (word,1)
            text.flatMap(new Tokenizer())
            // group by the tuple field "0" and sum up tuple field "1"
                    .keyBy(0)
                    .window(TumblingProcessingTimeWindows.of(Time.seconds(3)))

If I want to implement window wordcount.

Q1:Is there only one key in each window or multiple keys?

Q2:For the functions in the window, I only use simple sum++ or need to handle the sum of multiple keys through the hashmap in the window like Apache Storm.

Thank you for your help.

Answer 1

Even if there are actually multiple keys per window, each call to your process/reduce/sum/aggregate function is made with elements with the same key.

In your example you can just use sum and Flink will take care of everything:

text.flatMap(new Tokenizer())
      .keyBy(0)
      .window(TumblingProcessingTimeWindows.of(Time.seconds(3)))
      .sum(X)

If you chose to go with a reduce instead...

text.flatMap(new Tokenizer())
      .keyBy(0)
      .window(TumblingProcessingTimeWindows.of(Time.seconds(3)))
      .reduce(new ReduceFunction<Tuple2<String, Integer>>(){
            @Override
            public Tuple2<String, Integer> reduce(final Tuple2<String, Integer> first, final Tuple2<String, Integer> second) {
                  (... do something with the guarantee that first[0] == second[0] (same key) ...)
            }
      });