Spark Streaming与Flink进行实时数据处理方案对比

import org.apache.spark.sql.{SparkSession, DataFrame}
import org.apache.spark.sql.functions._

object SparkStreamingApp {
  def main(args: Array[String]): Unit = {
    val spark = SparkSession.builder()
      .appName("SparkStreamingCount")
      .getOrCreate()

    // 从 Kafka 读取数据
    val df: DataFrame = spark.readStream
      .format("kafka")
      .option("kafka.bootstrap.servers", "broker1:9092,broker2:9092")
      .option("subscribe", "access_logs")
      .load()

    // 假设 value = JSON，包含 timestamp 字段
    val logs = df.selectExpr("CAST(value AS STRING)")
      .select(from_json(col("value"), schemaOf[AccessLog]).as("data"))
      .select("data.timestamp")

    // 按分钟窗口聚合
    val result = logs
      .withColumn("eventTime", to_timestamp(col("timestamp")))
      .groupBy(window(col("eventTime"), "1 minute"))
      .count()

    val query = result.writeStream
      .outputMode("update")
      .format("console")
      .option("truncate", false)
      .trigger(processingTime = "30 seconds")
      .start()

    query.awaitTermination()
  }
}

spark {
  streaming.backpressure.enabled = true
  streaming.kafka.maxRatePerPartition = 10000
}

public class FlinkStreamingApp {
    public static void main(String[] args) throws Exception {
        StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();
        env.enableCheckpointing(60000); // 60s
        env.setStateBackend(new RocksDBStateBackend("hdfs://namenode:8020/flink/checkpoints", true));

        // Kafka Source
        Properties props = new Properties();
        props.setProperty("bootstrap.servers", "broker1:9092,broker2:9092");
        props.setProperty("group.id", "flink-group");

        DataStream<String> stream = env
            .addSource(new FlinkKafkaConsumer<>(
                "access_logs",
                new SimpleStringSchema(),
                props
            ));

        // 解析 JSON 并提取时间戳
        DataStream<AccessLog> logs = stream
            .map(json -> parseJson(json, AccessLog.class))
            .assignTimestampsAndWatermarks(
                WatermarkStrategy
                    .<AccessLog>forBoundedOutOfOrderness(Duration.ofSeconds(5))
                    .withTimestampAssigner((log, ts) -> log.getTimestamp())
            );

        // 按分钟窗口统计
        logs
          .keyBy(log -> "all")
          .window(TumblingEventTimeWindows.of(Time.minutes(1)))
          .process(new ProcessWindowFunction<AccessLog, Tuple2<String, Long>, String, TimeWindow>() {
              @Override
              public void process(String key, Context ctx, Iterable<AccessLog> elements, Collector<Tuple2<String, Long>> out) {
                  long count = StreamSupport.stream(elements.spliterator(), false).count();
                  out.collect(new Tuple2<>(ctx.window().toString(), count));
              }
          })
          .print();

        env.execute("FlinkStreamingCount");
    }
}