spark通过jdbc读取数据库的并行

代码如下:

    val conf = new SparkConf().setAppName("testMysqlToHiveJdbc")
                                           .setMaster("local")
    val spark = SparkSession.builder()
      .config(conf)
      .enableHiveSupport()
      .getOrCreate()
    ////定义Propertites,确定链接MySQL的参数
    val mysqlProperties = new java.util.Properties()
    //MySQL的jdbc链接
    val mysqlConnectionUrl = "jdbc:mysql://localhost:3306/rest"
    //定义检索语句,用于MySQL链接
    val mysqlTableName = """(select t.*,
    case when id<4000000 and id >=0 then 1
            when id<8000000 and id >=4000000 then 2
            when id<12000000 and id >=8000000 then 3
            when id<16000000 and id >=12000000 then 4
            when id<20000000 and id >=16000000 then 5
    else 6 end par
         from usppa_twitter_data t) tt"""
    //    val mysqlTableName = "usppa_twitter_data"
    mysqlProperties.put("driver","com.mysql.jdbc.Driver")   //确定driver
    mysqlProperties.put("user","root")          //用户名
    mysqlProperties.put("password","1234")      //密码
    mysqlProperties.put("fetchsize","10000")     //批次取数数量
    mysqlProperties.put("lowerBound","1")        //确定分区
    mysqlProperties.put("upperBound","7")           //确定分区
    mysqlProperties.put("numPartitions","6")        //分区数量
    mysqlProperties.put("partitionColumn","par")    //分区字段

    //读取数据
    val re = spark.read.jdbc(mysqlConnectionUrl, 
                   mysqlTableName,mysqlProperties)
    //写入Hive表中
    re.toDF().write.mode("overwrite").saveAsTable("testwarehouse.testtt")                            
View Code


代码中,lowerbound和upperbound有两种情况需要考虑。

1) 分区字段值可以穷举出来,如年份。

  引用外网:https://www.percona.com/blog/2016/08/17/apache-spark-makes-slow-mysql-queries-10x-faster/

  如下,lowerbound和upperbound会按照年份进行数据分区,这里的分区指的是并行的executors。

  

val jdbcDF = spark.read.format("jdbc").options(
     |   Map("url" ->  "jdbc:mysql://localhost:3306/ontime?user=root&password=mysql",
     |   "dbtable" -> "ontime.ontime_sm",
     |   "fetchSize" -> "10000",
     |   "partitionColumn" -> "yeard", "lowerBound" -> "1988", "upperBound" -> "2015", "numPartitions" -> "48"
     |   )).load()
View Code

  

  分区后,SQL会拆分成多个SQL:
  
2)分区字段不固定,如自动增长的ip,这时候lowerbound和upperbound在id数值之间,分区是一个估算值
  容易产生问题,每个executor的数据分布不均,导致OOM,源码带看。
  使用方式如下:  
CREATE OR REPLACE TEMPORARY VIEW ontime
USING org.apache.spark.sql.jdbc
OPTIONS (
  url  "jdbc:mysql://127.0.0.1:3306/ontime?user=root&password=",
  dbtable "ontime.ontime",
  fetchSize "1000",
  partitionColumn "id", lowerBound "1", upperBound "162668934", numPartitions "128"
);
View Code

 

 

 

posted @ 2019-06-16 18:20  Wind_LPH  阅读(2726)  评论(0编辑  收藏  举报