1. 技术文章
  2. How to read from hbase using spark

How to read from hbase using spark

由 ﹥＞﹥吖頭↗ 提交于 2019-11-26 21:30:34

A Basic Example to Read the HBase data using Spark (Scala), You can also wrtie this in Java : 

import org.apache.hadoop.hbase.client.{HBaseAdmin, Result}
import org.apache.hadoop.hbase.{ HBaseConfiguration, HTableDescriptor }
import org.apache.hadoop.hbase.mapreduce.TableInputFormat
import org.apache.hadoop.hbase.io.ImmutableBytesWritable

import org.apache.spark._

object HBaseRead {
  def main(args: Array[String]) {
    val sparkConf = new SparkConf().setAppName("HBaseRead").setMaster("local[2]")
    val sc = new SparkContext(sparkConf)
    val conf = HBaseConfiguration.create()
    val tableName = "table1"

    System.setProperty("user.name", "hdfs")
    System.setProperty("HADOOP_USER_NAME", "hdfs")
    conf.set("hbase.master", "localhost:60000")
    conf.setInt("timeout", 120000)
    conf.set("hbase.zookeeper.quorum", "localhost")
    conf.set("zookeeper.znode.parent", "/hbase-unsecure")
    conf.set(TableInputFormat.INPUT_TABLE, tableName)

    val admin = new HBaseAdmin(conf)
    if (!admin.isTableAvailable(tableName)) {
      val tableDesc = new HTableDescriptor(tableName)
      admin.createTable(tableDesc)
    }

    val hBaseRDD = sc.newAPIHadoopRDD(conf, classOf[TableInputFormat], classOf[ImmutableBytesWritable], classOf[Result])
    println("Number of Records found : " + hBaseRDD.count())
    sc.stop()
  }
}

UPDATED -2016

As of Spark 1.0.x+, Now you can use Spark-HBase Connector also :

Maven Dependency to Include : 

<dependency>
  <groupId>it.nerdammer.bigdata</groupId>
  <artifactId>spark-hbase-connector_2.10</artifactId>
  <version>1.0.3</version> // Version can be changed as per your Spark version, I am using Spark 1.6.x
</dependency>

And find a below sample code for the same : 

import org.apache.spark._
import it.nerdammer.spark.hbase._

object HBaseRead extends App {
    val sparkConf = new SparkConf().setAppName("Spark-HBase").setMaster("local[4]")
    sparkConf.set("spark.hbase.host", "<YourHostnameOnly>") //e.g. 192.168.1.1 or localhost or your hostanme
    val sc = new SparkContext(sparkConf)

    // For Example If you have an HBase Table as 'Document' with ColumnFamily 'SMPL' and qualifier as 'DocID, Title' then:

    val docRdd = sc.hbaseTable[(Option[String], Option[String])]("Document")
    .select("DocID", "Title").inColumnFamily("SMPL")

    println("Number of Records found : " + docRdd .count())
}

UPDATED - 2017

As of Spark 1.6.x+, Now you can use SHC Connector also (Hortonworks or HDP users) :

Maven Dependency to Include : 

    <dependency>
        <groupId>com.hortonworks</groupId>
        <artifactId>shc</artifactId>
        <version>1.0.0-2.0-s_2.11</version> // Version depends on the Spark version and is supported upto Spark 2.x
    </dependency>

The Main advantage of using this connector is that it have flexibility in the Schema definition and doesn't need Hardcoded params just like in nerdammer/spark-hbase-connector. Also remember that it supports Spark 2.x so this connector is pretty much flexible and provides end-to-end support in Issues and PRs.

Find the below repository path for the latest readme and samples :

Hortonworks Spark HBase Connector

You can also convert this RDD's to DataFrames and run SQL over it or You can map these Dataset or DataFrames to user defined Java Pojo's or Case classes. It works brilliant.

Please comment below if you need anything else.

I prefer to read from hbase and do the json manipulation all in spark.
Spark provides JavaSparkContext.newAPIHadoopRDD function to read data from hadoop storage, including HBase. You will have to provide the HBase configuration, table name, and scan in the configuration parameter and table input format and it's key-value

You can use table input format class and it's job parameter to provide the table name and scan configuration

example:

conf.set(TableInputFormat.INPUT_TABLE, "tablename");
JavaPairRDD<ImmutableBytesWritable, Result> data = 
jsc.newAPIHadoopRDD(conf, TableInputFormat.class,ImmutableBytesWritable.class, Result.class);

then you can do the json manipulation in spark. Since spark can do recalculation when the memory is full, it will only load the data needed for the recalculation part (cmiiw) so you don't have to worry about the data size

Zhang Kan

just to add a comment on how to add scan:

TableInputFormat has the following attributes:

  1. SCAN_ROW_START
  2. SCAN_ROW_STOP
conf.set(TableInputFormat.SCAN_ROW_START, "startrowkey");
conf.set(TableInputFormat.SCAN_ROW_STOP, "stoprowkey");

Since the question is not new, there are a few other alternatives for now:

I do not know much about the first project, but it looks like it does not support Spark 2.x. However, it has a rich support at the RDD level for Spark 1.6.x.

Spark-on-HBase, on the other hand, has branches for Spark 2.0 and upcoming Spark 2.1. This project is very promising since it is focused on Dataset/DataFrame APIs. Under the hood, it implements the standard Spark Datasource API and leverages the Spark Catalyst engine for query optimization. The developers claim here that it is capable of partition pruning, column pruning, predicate pushdown and achieving data locality.

A simple example, which uses the com.hortonworks:shc:1.0.0-2.0-s_2.11 artifact from this repo and Spark 2.0.2, is presented next: 

case class Record(col0: Int, col1: Int, col2: Boolean)

val spark = SparkSession
  .builder()
  .appName("Spark HBase Example")
  .master("local[4]")
  .getOrCreate()

def catalog =
  s"""{
      |"table":{"namespace":"default", "name":"table1"},
      |"rowkey":"key",
      |"columns":{
      |"col0":{"cf":"rowkey", "col":"key", "type":"int"},
      |"col1":{"cf":"cf1", "col":"col1", "type":"int"},
      |"col2":{"cf":"cf2", "col":"col2", "type":"boolean"}
      |}
      |}""".stripMargin

val artificialData = (0 to 100).map(number => Record(number, number, number % 2 == 0))

// write
spark
  .createDataFrame(artificialData)
  .write
  .option(HBaseTableCatalog.tableCatalog, catalog)
  .option(HBaseTableCatalog.newTable, "5")
  .format("org.apache.spark.sql.execution.datasources.hbase")
  .save()

// read
val df = spark
  .read
  .option(HBaseTableCatalog.tableCatalog, catalog)
  .format("org.apache.spark.sql.execution.datasources.hbase")
  .load()

df.count()
标签
hbase
apache-spark
rdd
易学教程内所有资源均来自网络或用户发布的内容，如有违反法律规定的内容欢迎反馈！
该文章没有解决你所遇到的问题?点击提问,说说你的问题,让更多的人一起探讨吧!