使用spark-sql的Hbase

Question

我在hbase中有一個名爲「sample」的表。我需要使用Apache spark-sql查詢來查詢表。 有什麼方法可以使用Apache spark-sql查詢來讀取hbase數據嗎？

Answer 1

星火SQL是在內存中的查詢引擎，利用星火SQL對HBase的桌子上執行一些查詢操作，你需要

1. 使用星火從HBase的獲取數據，並創建星火RDD

   SparkConf sparkConf = new SparkConf(); 
   sparkConf.setAppName("SparkApp"); 
   sparkConf.setMaster("local[*]"); 
   
   JavaSparkContext javaSparkContext = new JavaSparkContext(sparkConf); 
   
   Configuration config = HBaseConfiguration.create(); 
   config.addResource(new Path("/etc/hbase/hbase-site.xml")); 
   config.addResource(new Path("/etc/hadoop/core-site.xml")); 
   config.set(TableInputFormat.INPUT_TABLE, "sample"); 
   
   JavaPairRDD<ImmutableBytesWritable, Result> hbaseRDD = javaSparkContext.newAPIHadoopRDD(hbaseConfig, TableInputFormat.class, ImmutableBytesWritable.class, Result.class); 
   
   JavaRDD<StudentBean> sampleRDD = hbaseRDD.map(new Function<Tuple2<ImmutableBytesWritable,Result>, StudentBean >() { 
       private static final long serialVersionUID = -2021713021648730786L; 
       public StudentBean call(Tuple2<ImmutableBytesWritable, Result> tuple) { 
        StudentBean bean = new StudentBean (); 
        Result result = tuple._2; 
        bean.setRowKey(rowKey); 
        bean.setFirstName(Bytes.toString(result.getValue(Bytes.toBytes("details"), Bytes.toBytes("firstName")))); 
        bean.setLastName(Bytes.toString(result.getValue(Bytes.toBytes("details"), Bytes.toBytes("lastName")))); 
        bean.setBranch(Bytes.toString(result.getValue(Bytes.toBytes("details"), Bytes.toBytes("branch")))); 
        bean.setEmailId(Bytes.toString(result.getValue(Bytes.toBytes("details"), Bytes.toBytes("emailId")))); 
        return bean; 
       } 
   }); 
   

2. 通過使用此RDD創建數據框對象，並使用一些臨時表的名字註冊這一點，那麼你就可以執行查詢

   DataFrame schema = sqlContext.createDataFrame(sampleRDD, StudentBean.class); 
   schema.registerTempTable("spark_sql_temp_table"); 
   
   DataFrame schemaRDD = sqlContext.sql("YOUR_QUERY_GOES_HERE"); 
   
   JavaRDD<StudentBean> result = schemaRDD.toJavaRDD().map(new Function<Row, StudentBean>() { 
   
       private static final long serialVersionUID = -2558736294883522519L; 
   
       public StudentBean call(Row row) throws Exception { 
        StudentBean bean = new StudentBean(); 
        // Do the mapping stuff here 
        return bean; 
       } 
   });