Insert Spark dataframe into hbase

Question

I have a dataframe and I want to insert it into hbase. I follow this documenation .

This is how my dataframe look like:

 --------------------
|id | name | address |
|--------------------|
|23 |marry |france   |
|--------------------|
|87 |zied  |italie   |
 --------------------

I create a hbase table using this code:

val tableName = "two"
val conf = HBaseConfiguration.create()
if(!admin.isTableAvailable(tableName)) {
          print("-----------------------------------------------------------------------------------------------------------")
          val tableDesc = new HTableDescriptor(tableName)
          tableDesc.addFamily(new HColumnDescriptor("z1".getBytes()))
          admin.createTable(tableDesc)
        }else{
          print("Table already exists!!--------------------------------------------------------------------------------------")
        }

And now how may I insert this dataframe into hbase ?

In another example I succeed to insert into hbase using this code:

val myTable = new HTable(conf, tableName)
    for (i <- 0 to 1000) {
      var p = new Put(Bytes.toBytes(""+i))
      p.add("z1".getBytes(), "name".getBytes(), Bytes.toBytes(""+(i*5)))
      p.add("z1".getBytes(), "age".getBytes(), Bytes.toBytes("2017-04-20"))
      p.add("z2".getBytes(), "job".getBytes(), Bytes.toBytes(""+i))
      p.add("z2".getBytes(), "salary".getBytes(), Bytes.toBytes(""+i))
      myTable.put(p)
    }
    myTable.flushCommits()

But now I am stuck, how to insert each record of my dataframe into my hbase table.

Thank you for your time and attention

Answer 1

Below is a full example using the spark hbase connector from Hortonworks available inMaven .

This example shows

• how to check if HBase table is existing
• create HBase table if not existing
• Insert DataFrame into HBase table

import org.apache.hadoop.hbase.client.{ColumnFamilyDescriptorBuilder, ConnectionFactory, TableDescriptorBuilder}
import org.apache.hadoop.hbase.{HBaseConfiguration, TableName}
import org.apache.spark.sql.SparkSession
import org.apache.spark.sql.execution.datasources.hbase.HBaseTableCatalog

object Main extends App {

  case class Employee(key: String, fName: String, lName: String, mName: String,
                      addressLine: String, city: String, state: String, zipCode: String)

  // as pre-requisites the table 'employee' with column families 'person' and 'address' should exist
  val tableNameString = "default:employee"
  val colFamilyPString = "person"
  val colFamilyAString = "address"
  val tableName = TableName.valueOf(tableNameString)
  val colFamilyP = colFamilyPString.getBytes
  val colFamilyA = colFamilyAString.getBytes

  val hBaseConf = HBaseConfiguration.create()
  val connection = ConnectionFactory.createConnection(hBaseConf);
  val admin = connection.getAdmin();

  println("Check if table 'employee' exists:")
  val tableExistsCheck: Boolean = admin.tableExists(tableName)
  println(s"Table " + tableName.toString + " exists? " + tableExistsCheck)

  if(tableExistsCheck == false) {
    println("Create Table employee with column families 'person' and 'address'")
    val colFamilyBuild1 = ColumnFamilyDescriptorBuilder.newBuilder(colFamilyP).build()
    val colFamilyBuild2 = ColumnFamilyDescriptorBuilder.newBuilder(colFamilyA).build()
    val tableDescriptorBuild = TableDescriptorBuilder.newBuilder(tableName)
      .setColumnFamily(colFamilyBuild1)
      .setColumnFamily(colFamilyBuild2)
      .build()
    admin.createTable(tableDescriptorBuild)
  }

  // define schema for the dataframe that should be loaded into HBase
  def catalog =
    s"""{
       |"table":{"namespace":"default","name":"employee"},
       |"rowkey":"key",
       |"columns":{
       |"key":{"cf":"rowkey","col":"key","type":"string"},
       |"fName":{"cf":"person","col":"firstName","type":"string"},
       |"lName":{"cf":"person","col":"lastName","type":"string"},
       |"mName":{"cf":"person","col":"middleName","type":"string"},
       |"addressLine":{"cf":"address","col":"addressLine","type":"string"},
       |"city":{"cf":"address","col":"city","type":"string"},
       |"state":{"cf":"address","col":"state","type":"string"},
       |"zipCode":{"cf":"address","col":"zipCode","type":"string"}
       |}
       |}""".stripMargin

  // define some test data
  val data = Seq(
    Employee("1","Horst","Hans","A","12main","NYC","NY","123"),
    Employee("2","Joe","Bill","B","1337ave","LA","CA","456"),
    Employee("3","Mohammed","Mohammed","C","1Apple","SanFran","CA","678")
  )

  // create SparkSession
  val spark: SparkSession = SparkSession.builder()
    .master("local[*]")
    .appName("HBaseConnector")
    .getOrCreate()

  // serialize data
  import spark.implicits._
  val df = spark.sparkContext.parallelize(data).toDF

  // write dataframe into HBase
  df.write.options(
    Map(HBaseTableCatalog.tableCatalog -> catalog, HBaseTableCatalog.newTable -> "3")) // create 3 regions
    .format("org.apache.spark.sql.execution.datasources.hbase")
    .save()

}

This worked for me while I had the relevant site-xmls ("core-site.xml", "hbase-site.xml", "hdfs-site.xml") available in my resources .

Answer 2

using answer for code formatting purposes Doc tells:

sc.parallelize(data).toDF.write.options(
 Map(HBaseTableCatalog.tableCatalog -> catalog, HBaseTableCatalog.newTable -> "5"))
 .format("org.apache.hadoop.hbase.spark ")
 .save()

where sc.parallelize(data).toDF is your DataFrame. Doc example turns scala collection to dataframe using sc.parallelize(data).toDF

You already have your DataFrame, just try to call

yourDataFrame.write.options(
     Map(HBaseTableCatalog.tableCatalog -> catalog, HBaseTableCatalog.newTable -> "5"))
     .format("org.apache.hadoop.hbase.spark ")
     .save()

And it should work. Doc is pretty clear...

UPD

Given a DataFrame with specified schema, above will create an HBase table with 5 regions and save the DataFrame inside. Note that if HBaseTableCatalog.newTable is not specified, the table has to be pre-created.

It's about data partitioning. Each HBase table can have 1...X regions. You should carefully pick number of regions. Low regions number is bad. High region numbers is also bad.

Answer 3

An alternate is to look at rdd.saveAsNewAPIHadoopDataset, to insert the data into the hbase table.

def main(args: Array[String]): Unit = {

    val spark = SparkSession.builder().appName("sparkToHive").enableHiveSupport().getOrCreate()
    import spark.implicits._

    val config = HBaseConfiguration.create()
    config.set("hbase.zookeeper.quorum", "ip's")
    config.set("hbase.zookeeper.property.clientPort","2181")
    config.set(TableInputFormat.INPUT_TABLE, "tableName")

    val newAPIJobConfiguration1 = Job.getInstance(config)
    newAPIJobConfiguration1.getConfiguration().set(TableOutputFormat.OUTPUT_TABLE, "tableName")
    newAPIJobConfiguration1.setOutputFormatClass(classOf[TableOutputFormat[ImmutableBytesWritable]])

    val df: DataFrame  = Seq(("foo", "1", "foo1"), ("bar", "2", "bar1")).toDF("key", "value1", "value2")

    val hbasePuts= df.rdd.map((row: Row) => {
      val  put = new Put(Bytes.toBytes(row.getString(0)))
      put.addColumn(Bytes.toBytes("cf1"), Bytes.toBytes("value1"), Bytes.toBytes(row.getString(1)))
      put.addColumn(Bytes.toBytes("cf2"), Bytes.toBytes("value2"), Bytes.toBytes(row.getString(2)))
      (new ImmutableBytesWritable(), put)
    })

    hbasePuts.saveAsNewAPIHadoopDataset(newAPIJobConfiguration1.getConfiguration())
    }

Ref : https://sparkkb.wordpress.com/2015/05/04/save-javardd-to-hbase-using-saveasnewapihadoopdataset-spark-api-java-coding/