如何 运行 多重 Spark Cassandra 查询

Question

我需要运行下面这样的任务。不知何故，我错过了一点。我知道，我不能像这样使用 javasparkcontext 并传递 javafunctions，因为存在序列化问题。

我需要 运行 多个大小为 cartesian.size() 的 cassandra 查询。有什么建议吗？

JavaSparkContext jsc = new JavaSparkContext(conf);
    JavaRDD<DateTime> dateTimeJavaRDD = jsc.parallelize(dateTimes); //List<DateTime>
    JavaRDD<Integer> virtualPartitionJavaRDD = jsc.parallelize(virtualPartitions); //List<Integer>
    JavaPairRDD<DateTime, Integer> cartesian = dateTimeJavaRDD.cartesian(virtualPartitionJavaRDD);

    long c = cartesian.map(new Function<Tuple2<DateTime, Integer>, Long>() {
        @Override
        public Long call(Tuple2<DateTime, Integer> tuple2) throws Exception {
            return javaFunctions(jsc).cassandraTable("keyspace", "table").where("p1 = ? and  p2 = ?", tuple2._1(), tuple2._2()).count();
        }
    }).reduce((a,b) -> a + b);


    System.out.println("TOTAL ROW COUNT IS: " + c);

Answer 1

正确的解决方案应该是在您的数据和 Casasndra table 之间执行连接。有joinWithCassandraTable function that is doing what you need - you just generate RDD of Tuple2 that contains values for p1 & p2, and then call joinWithCassandra table, something like this (not tested, adopted from my example here):

JavaRDD<Tuple2<Integer, Integer>> trdd = cartesian.map(new Function<Tuple2<DateTime, Integer>, Tuple2<Integer, Integer>>() {
        @Override
        public Tuple2<Integer, Integer> call(Tuple2<DateTime, Integer> tuple2) throws Exception {
            return new Tuple2<Integer, Integer>(tuple2._1(), tuple2._2());
        }
    });
CassandraJavaPairRDD<Tuple2<Integer, Integer>, Tuple2<Integer, String>> joinedRDD =
     trdd.joinWithCassandraTable("test", "jtest",
     someColumns("p1", "p2"), someColumns("p1", "p2"),
     mapRowToTuple(Integer.class, String.class), mapTupleToRow(Integer.class));
// perform counting here...