Spark如何批量存取HBase

这篇文章将为大家详细讲解有关Spark如何批量存取HBase ，小编觉得挺实用的，因此分享给大家做个参考，希望大家阅读完这篇文章后可以有所收获。

FileAna.scala

object FileAna {

  //  val conf: Configuration = HBaseConfiguration.create()

  val hdfsPath = "hdfs://master:9000"
  val hdfs = FileSystem.get(new URI(hdfsPath), new Configuration())

  def main(args: Array[String]) {
    val conf = new SparkConf().setAppName("FileAna").setMaster("spark://master:7077").
      set("spark.driver.host", "192.168.1.127").
      setJars(List("/home/pang/woozoomws/spark-service.jar",
        "/home/pang/woozoomws/spark-service/lib/hbase/hbase-common-1.2.2.jar",
        "/home/pang/woozoomws/spark-service/lib/hbase/hbase-client-1.2.2.jar",
        "/home/pang/woozoomws/spark-service/lib/hbase/hbase-protocol-1.2.2.jar",
        "/home/pang/woozoomws/spark-service/lib/hbase/htrace-core-3.1.0-incubating.jar",
        "/home/pang/woozoomws/spark-service/lib/hbase/hbase-server-1.2.2.jar",
        "/home/pang/woozoomws/spark-service/lib/hbase/metrics-core-2.2.0.jar"))
    val sc = new SparkContext(conf)
    val rdd = sc.textFile("hdfs://master:9000/woozoom/msgfile.txt")
    val rdd2 = rdd.map(x => convertToHbase(anaMavlink(x)))

    val hbaseConf = HBaseConfiguration.create()
    hbaseConf.addResource("/home/hadoop/software/hbase-1.2.2/conf/hbase-site.xml");

    val jobConf = new JobConf(hbaseConf, this.getClass)
    jobConf.setOutputFormat(classOf[TableOutputFormat])
    jobConf.set(TableOutputFormat.OUTPUT_TABLE, "MissionItem")

    rdd2.saveAsHadoopDataset(jobConf)

    sc.stop()
  }

  def convertScanToString(scan: Scan) = {
    val proto = ProtobufUtil.toScan(scan)
    Base64.encodeBytes(proto.toByteArray)
  }

  def convertToHbase(msg: MAVLinkMessage) = {
    val p = new Put(Bytes.toBytes(UUID.randomUUID().toString()))
    if (msg.isInstanceOf[msg_mission_item]) {
      val missionItem = msg.asInstanceOf[msg_mission_item]
      p.addColumn(Bytes.toBytes("data"), Bytes.toBytes("x"), Bytes.toBytes(missionItem.x))
      p.addColumn(Bytes.toBytes("data"), Bytes.toBytes("y"), Bytes.toBytes(missionItem.y))
      p.addColumn(Bytes.toBytes("data"), Bytes.toBytes("z"), Bytes.toBytes(missionItem.z))
    }
    (new ImmutableBytesWritable, p)
  }

  val anaMavlink = (str: String) => {
    val bytes = ByteAndHex.hexStringToBytes(str)
    QuickParser.parse(bytes).unpack()
  }
}

ReadHBase.scala

object ReadHBase {

  //  val conf: Configuration = HBaseConfiguration.create()

  val hdfsPath = "hdfs://master:9000"
  val hdfs = FileSystem.get(new URI(hdfsPath), new Configuration())

  def main(args: Array[String]) {
    val conf = new SparkConf().setAppName("FileAna").setMaster("spark://master:7077").
      set("spark.driver.host", "192.168.1.127").
      setJars(List("/home/pang/woozoomws/spark-service.jar",
        "/home/pang/woozoomws/spark-service/lib/hbase/hbase-common-1.2.2.jar",
        "/home/pang/woozoomws/spark-service/lib/hbase/hbase-client-1.2.2.jar",
        "/home/pang/woozoomws/spark-service/lib/hbase/hbase-protocol-1.2.2.jar",
        "/home/pang/woozoomws/spark-service/lib/hbase/htrace-core-3.1.0-incubating.jar",
        "/home/pang/woozoomws/spark-service/lib/hbase/hbase-server-1.2.2.jar",
        "/home/pang/woozoomws/spark-service/lib/hbase/metrics-core-2.2.0.jar"))
    val sc = new SparkContext(conf)

    val hbaseConf = HBaseConfiguration.create()
    hbaseConf.addResource("/home/hadoop/software/hbase-1.2.2/conf/hbase-site.xml");

    hbaseConf.set(TableInputFormat.INPUT_TABLE, "MissionItem")
    val scan = new Scan()
    hbaseConf.set(TableInputFormat.SCAN, convertScanToString(scan))
    val readRDD = sc.newAPIHadoopRDD(hbaseConf, classOf[TableInputFormat],
      classOf[org.apache.hadoop.hbase.io.ImmutableBytesWritable],
      classOf[org.apache.hadoop.hbase.client.Result])

    val count = readRDD.count()
    println("Mission Item Count:" + count)

    sc.stop()
  }

  def convertScanToString(scan: Scan) = {
    val proto = ProtobufUtil.toScan(scan)
    Base64.encodeBytes(proto.toByteArray)
  }
}

关于“Spark如何批量存取HBase ”这篇文章就分享到这里了，希望以上内容可以对大家有一定的帮助，使各位可以学到更多知识，如果觉得文章不错，请把它分享出去让更多的人看到。