Spark JDBC方式连接MySQL数据库

Spark JDBC方式连接MySQL数据库

• 一、JDBC connection properties（属性名称和含义）
• 二、spark jdbc read MySQL
• 三、jdbc(url: String, table: String, properties: Properties): DataFrame
• 四、jdbc(url: String, table: String, columnName: String, lowerBound: Long, upperBound: Long, numPartitions: Int, connectionProperties: Properties): DataFrame
• 四、jdbc(url: String, table: String, predicates: Array[String], connectionProperties: Properties): DataFrame predicates: Condition in the WHERE clause for each partition.
• 五、spark jdbc write MySQL

一、JDBC connection properties（属性名称和含义）

url：要连接的JDBC URL。列如：jdbc:mysql://ip:3306

dbtable:应该读取的JDBC表。可以使用括号中的子查询代替完整表。

driver:用于连接到此URL的JDBC驱动程序的类名,列如：com.mysql.jdbc.Driver

partitionColumn, lowerBound, upperBound, numPartitions：

这些options仅适用于read数据。这些options必须同时被指定。他们描述，如何从多个workers并行读取数据时，分割表。

partitionColumn：必须是表中的数字列。

lowerBound和upperBound仅用于决定分区的大小，而不是用于过滤表中的行。

表中的所有行将被分割并返回。

fetchsize：仅适用于read数据。JDBC提取大小，用于确定每次获取的行数。这可以帮助JDBC驱动程序调优性能，这些驱动程序默认具有较低的提取大小（例如，Oracle每次提取10行）。

batchsize：仅适用于write数据。JDBC批量大小，用于确定每次insert的行数。

这可以帮助JDBC驱动程序调优性能。默认为1000。

isolationLevel：仅适用于write数据。事务隔离级别，适用于当前连接。它可以是一个NONE，READ_COMMITTED，READ_UNCOMMITTED，REPEATABLE_READ，或SERIALIZABLE，对应于由JDBC的连接对象定义，缺省值为标准事务隔离级别READ_UNCOMMITTED。请参阅文档java.sql.Connection。

truncate：仅适用于write数据。当SaveMode.Overwrite启用时，此选项会truncate在MySQL中的表，而不是删除，再重建其现有的表。这可以更有效，并且防止表元数据（例如，索引）被去除。但是，在某些情况下，例如当新数据具有不同的模式时，它将无法工作。它默认为false。

createTableOptions：仅适用于write数据。此选项允许在创建表（例如CREATE TABLE t (name string) ENGINE=InnoDB.）时设置特定的数据库表和分区选项。

二、spark jdbc read MySQL

val jdbcDF11 = spark.read.format("jdbc")
      .option("driver", "com.mysql.jdbc.Driver")
      .option("url", "jdbc:mysql://ip:3306")
      .option("dbtable", "db.user_test")
      .option("user", "test")
      .option("password", "123456")
      .option("fetchsize", "3")
      .load()
jdbcDF11.show

val jdbcDF12 = spark.read.format("jdbc").options(
      Map(
        "driver" -> "com.mysql.jdbc.Driver",
        "url" -> "jdbc:mysql://ip:3306",
        "dbtable" -> "db.user_test",
        "user" -> "test",
        "password" -> "123456",
        "fetchsize" -> "3")).load()
jdbcDF12.show

三、jdbc(url: String, table: String, properties: Properties): DataFrame

import java.util.Properties

// jdbc(url: String, table: String, properties: Properties): DataFrame

val readConnProperties1 = new Properties()
readConnProperties1.put("driver", "com.mysql.jdbc.Driver")
readConnProperties1.put("user", "test")
readConnProperties1.put("password", "123456")
readConnProperties1.put("fetchsize", "3")

val jdbcDF1 = spark.read.jdbc(
  "jdbc:mysql://ip:3306",
  "db.user_test",
  readConnProperties1)

jdbcDF1.show
+---+------+---+
|uid|gender|age|
+---+------+---+
|  2|     2| 20|
|  3|     1| 30|
|  4|     2| 40|
|  5|     1| 50|
|  6|     2| 60|
|  7|     1| 25|
|  8|     2| 35|
|  9|     1| 70|
| 10|     2| 80|
|  1|     1| 18|
+---+------+---+

//默认并行度为1
jdbcDF1.rdd.partitions.size
Int = 1

//-------------------------

// jdbc(url: String, table: String, properties: Properties): DataFrame

val readConnProperties4 = new Properties()
readConnProperties4.put("driver", "com.mysql.jdbc.Driver")
readConnProperties4.put("user", "test")
readConnProperties4.put("password", "123456")
readConnProperties4.put("fetchsize", "3")

val jdbcDF4 = spark.read.jdbc(
  "jdbc:mysql://ip:3306",
  "(select * from db.user_test where gender=1) t",  // 注意括号和表别名，必须得有，这里可以过滤数据
  readConnProperties4)

jdbcDF4.show
+---+------+---+
|uid|gender|age|
+---+------+---+
|  3|     1| 30|
|  5|     1| 50|
|  7|     1| 25|
|  9|     1| 70|
|  1|     1| 18|
+---+------+---+

四、jdbc(url: String, table: String, columnName: String, lowerBound: Long, upperBound: Long, numPartitions: Int, connectionProperties: Properties): DataFrame

import java.util.Properties

val readConnProperties2 = new Properties()
readConnProperties2.put("driver", "com.mysql.jdbc.Driver")
readConnProperties2.put("user", "test")
readConnProperties2.put("password", "123456")
readConnProperties2.put("fetchsize", "2")

val columnName = "uid"
val lowerBound = 1
val upperBound = 6
val numPartitions = 3

val jdbcDF2 = spark.read.jdbc(
  "jdbc:mysql://ip:3306",
  "db.user_test",
  columnName,
  lowerBound,
  upperBound,
  numPartitions,
  readConnProperties2)

jdbcDF2.show
+---+------+---+
|uid|gender|age|
+---+------+---+
|  2|     2| 20|
|  1|     1| 18|
|  3|     1| 30|
|  4|     2| 40|
|  5|     1| 50|
|  6|     2| 60|
|  7|     1| 25|
|  8|     2| 35|
|  9|     1| 70|
| 10|     2| 80|
+---+------+---+

// 并行度为3，对应于numPartitions
jdbcDF2.rdd.partitions.size
Int = 3

四、jdbc(url: String, table: String, predicates: Array[String], connectionProperties: Properties): DataFrame predicates: Condition in the WHERE clause for each partition.

import java.util.Properties

val readConnProperties3 = new Properties()
readConnProperties3.put("driver", "com.mysql.jdbc.Driver")
readConnProperties3.put("user", "test")
readConnProperties3.put("password", "123456")
readConnProperties3.put("fetchsize", "2")

val arr = Array(
  (1, 50),
  (2, 60))

// 此处的条件，既可以分割数据用作并行度，也可以过滤数据
val predicates = arr.map {
  case (gender, age) =>
    s" gender = $gender " + s" AND age < $age "
}

val predicates1 =
  Array(
    "2017-05-01" -> "2017-05-20",
    "2017-06-01" -> "2017-06-05").map {
      case (start, end) =>
        s"cast(create_time as date) >= date '$start' " + s"AND cast(create_time as date) <= date '$end'"
    }

val jdbcDF3 = spark.read.jdbc(
  "jdbc:mysql://ip:3306",
  "db.user_test",
  predicates,
  readConnProperties3)

jdbcDF3.show
+---+------+---+
|uid|gender|age|
+---+------+---+
|  3|     1| 30|
|  7|     1| 25|
|  1|     1| 18|
|  2|     2| 20|
|  4|     2| 40|
|  8|     2| 35|
+---+------+---+

// 并行度为2，对应arr数组中元素的个数
jdbcDF3.rdd.partitions.size
Int = 2

五、spark jdbc write MySQL

import spark.implicits._

val dataList: List[(Double, String, Double, Double, String, Double, Double, Double, Double)] = List(
  (0, "male", 37, 10, "no", 3, 18, 7, 4),
  (0, "female", 27, 4, "no", 4, 14, 6, 4),
  (0, "female", 32, 15, "yes", 1, 12, 1, 4),
  (0, "male", 57, 15, "yes", 5, 18, 6, 5),
  (0, "male", 22, 0.75, "no", 2, 17, 6, 3),
  (0, "female", 32, 1.5, "no", 2, 17, 5, 5),
  (0, "female", 22, 0.75, "no", 2, 12, 1, 3),
  (0, "male", 57, 15, "yes", 2, 14, 4, 4),
  (0, "female", 32, 15, "yes", 4, 16, 1, 2))

val colArray: Array[String] = Array("affairs", "gender", "age", "yearsmarried", "children", "religiousness", "education", "occupation", "rating")

val df = dataList.toDF(colArray: _*)

df.write.mode("overwrite").format("jdbc").options(
  Map(
    "driver" -> "com.mysql.jdbc.Driver",
    "url" -> "jdbc:mysql://ip:3306",
    "dbtable" -> "db.affairs",
    "user" -> "test",
    "password" -> "123456",
    "batchsize" -> "1000",
    "truncate" -> "true")).save()