spark到hbase

1）spark把数据写入到hbase需要用到：PairRddFunctions的saveAsHadoopDataset方法，这里用到了 implicit conversion，需要我们引入

import org.apache.spark.SparkContext._

2）spark写入hbase，实质是借用了org.apache.hadoop.hbase.mapreduce.TableInputFormat这个对象，用其内部的recorderWriter将数据写入hbase

同时，也借用了hadoop的JobConf，配置和写MR的配置方式一样

3）请看下面代码,这里使用sparksql从hive里面读出数据，经过处理，写入到hbase

package savehbase
 
import org.apache.hadoop.hbase.HBaseConfiguration
import org.apache.hadoop.hbase.client.Put
import org.apache.hadoop.hbase.io.ImmutableBytesWritable
import org.apache.hadoop.hbase.mapred.TableOutputFormat
import org.apache.hadoop.hbase.util.Bytes
import org.apache.hadoop.mapred.JobConf
import org.apache.spark.sql.hive.HiveContext
import org.apache.spark.{SparkConf, SparkContext}
 
object SaveHbase3 {
  def main(args: Array[String]): Unit = {
    val config = HBaseConfiguration.create()
    val jobConf = new JobConf(config)
    jobConf.setOutputFormat(classOf[TableOutputFormat])
    jobConf.set(TableOutputFormat.OUTPUT_TABLE,"table2")
 
 
    val sparkconf = new SparkConf().setAppName("ss").setMaster("local")
    val sc = new SparkContext(sparkconf)
    val hiveContext = new HiveContext(sc)
    hiveContext.setConf("spark.sql.shuffle.partitions","3")
 
    hiveContext.sql("select id,name from hiv3.tablea")
      .rdd.map(row => {
      val id = row(0).asInstanceOf[Int].toString
      val name = row(1).asInstanceOf[String]
 
      val put = new Put(Bytes.toBytes(id))
 
      put.addColumn(Bytes.toBytes("f1"), Bytes.toBytes("name"), Bytes.toBytes(name))
 
      (new ImmutableBytesWritable, put)
 
 
    }).saveAsHadoopDataset(jobConf)
 
 
  }
}

遇到的问题：

1. 错误:(26, 11) Unable to find encoder for type stored in a Dataset. Primitive types (Int, String, etc) and Product types (case classes) are supported by importing spark.implicits._ Support for serializing other types will be added in future releases.

2. .map(row => {

经过查看spark官方文档，对spark有了一条这样的描述。

Dataset is Spark SQL’s strongly-typed API for working with structured data, i.e. records with a known schema.

Datasets are lazy and structured query expressions are only triggered when an action is invoked. Internally, aDataset represents a logical plan that describes the computation query required to produce the data (for a givenSpark SQL session).

A Dataset is a result of executing a query expression against data storage like files, Hive tables or JDBC databases. The structured query expression can be described by a SQL query, a Column-based SQL expression or a Scala/Java lambda function. And that is why Dataset operations are available in three variants.

从这可以看出，要想对dataset进行操作，需要进行相应的encode操作。特别是官网给的例子

// No pre-defined encoders for Dataset[Map[K,V]], define explicitly
implicit val mapEncoder = org.apache.spark.sql.Encoders.kryo[Map[String, Any]]
// Primitive types and case classes can be also defined as
// implicit val stringIntMapEncoder: Encoder[Map[String, Any]] = ExpressionEncoder()
 
// row.getValuesMap[T] retrieves multiple columns at once into a Map[String, T]
teenagersDF.map(teenager => teenager.getValuesMap[Any](List("name", "age"))).collect()
// Array(Map("name" -> "Justin", "age" -> 19))

从这看出，要进行map操作，要先定义一个Encoder。。

这就增加了系统升级繁重的工作量了。为了更简单一些，幸运的dataset也提供了转化RDD的操作。因此只需要将之前dataframe.map

在中间修改为：dataframe.rdd.map即可。