JDBC source connector

Description

Read external data source data through JDBC.

Warn: for license compliance, you have to provide database driver yourself, copy to $SEATNUNNEL_HOME/lib/ directory in order to make them work.

e.g. If you use MySQL, should download and copy mysql-connector-java-xxx.jar to $SEATNUNNEL_HOME/lib/. For Spark/Flink, you should also copy it to $SPARK_HOME/jars/ or $FLINK_HOME/lib/.

Using Dependency

For Spark/Flink Engine

1. You need to ensure that the jdbc driver jar package has been placed in directory ${SEATUNNEL_HOME}/plugins/.

For SeaTunnel Zeta Engine

1. You need to ensure that the jdbc driver jar package has been placed in directory ${SEATUNNEL_HOME}/lib/.

Key features

• batch
• stream
• exactly-once
• column projection

supports query SQL and can achieve projection effect.

• parallelism
• support user-defined split
• support multiple table read

Options

driver [string]

The jdbc class name used to connect to the remote data source, if you use MySQL the value is com.mysql.cj.jdbc.Driver.

user [string]

userName

password [string]

password

url [string]

The URL of the JDBC connection. Refer to a case: jdbc:postgresql://localhost/test

query [string]

Query statement

compatible_mode [string]

The compatible mode of database, required when the database supports multiple compatible modes. For example, when using OceanBase database, you need to set it to ‘mysql’ or ‘oracle’.

connection_check_timeout_sec [int]

The time in seconds to wait for the database operation used to validate the connection to complete.

fetch_size [int]

For queries that return a large number of objects, you can configure the row fetch size used in the query to improve performance by reducing the number database hits required to satisfy the selection criteria. Zero means use jdbc default value.

properties

Additional connection configuration parameters,when properties and URL have the same parameters, the priority is determined by the
specific implementation of the driver. For example, in MySQL, properties take precedence over the URL.

table_path

The path to the full path of table, you can use this configuration instead of query.

examples:

• mysql: “testdb.table1”
• oracle: “test_schema.table1”
• sqlserver: “testdb.test_schema.table1”
• postgresql: “testdb.test_schema.table1”

table_list

The list of tables to be read, you can use this configuration instead of table_path

example

table_list = [
  {
    table_path = "testdb.table1"
  }
  {
    table_path = "testdb.table2"
    query = "select * from testdb.table2 where id > 100"
  }
]

where_condition

Common row filter conditions for all tables/queries, must start with where. for example where id > 100

common options

Source plugin common parameters, please refer to Source Common Options for details.

Parallel Reader

The JDBC Source connector supports parallel reading of data from tables. SeaTunnel will use certain rules to split the data in the table, which will be handed over to readers for reading. The number of readers is determined by the parallelism option.

Split Key Rules:

1. If partition_column is not null, It will be used to calculate split. The column must in Supported split data type.
2. If partition_column is null, seatunnel will read the schema from table and get the Primary Key and Unique Index. If there are more than one column in Primary Key and Unique Index, The first column which in the supported split data type will be used to split data. For example, the table have Primary Key(nn guid, name varchar), because guid id not in supported split data type, so the column name will be used to split data.

Supported split data type:

• String
• Number(int, bigint, decimal, …)
• Date

Options Related To Split

split.size

How many rows in one split, captured tables are split into multiple splits when read of table.

split.even-distribution.factor.lower-bound

Not recommended for use

The lower bound of the chunk key distribution factor. This factor is used to determine whether the table data is evenly distributed. If the distribution factor is calculated to be greater than or equal to this lower bound (i.e., (MAX(id) - MIN(id) + 1) / row count), the table chunks would be optimized for even distribution. Otherwise, if the distribution factor is less, the table will be considered as unevenly distributed and the sampling-based sharding strategy will be used if the estimated shard count exceeds the value specified by sample-sharding.threshold. The default value is 0.05.

split.even-distribution.factor.upper-bound

Not recommended for use

The upper bound of the chunk key distribution factor. This factor is used to determine whether the table data is evenly distributed. If the distribution factor is calculated to be less than or equal to this upper bound (i.e., (MAX(id) - MIN(id) + 1) / row count), the table chunks would be optimized for even distribution. Otherwise, if the distribution factor is greater, the table will be considered as unevenly distributed and the sampling-based sharding strategy will be used if the estimated shard count exceeds the value specified by sample-sharding.threshold. The default value is 100.0.

split.sample-sharding.threshold

This configuration specifies the threshold of estimated shard count to trigger the sample sharding strategy. When the distribution factor is outside the bounds specified by chunk-key.even-distribution.factor.upper-bound and chunk-key.even-distribution.factor.lower-bound, and the estimated shard count (calculated as approximate row count / chunk size) exceeds this threshold, the sample sharding strategy will be used. This can help to handle large datasets more efficiently. The default value is 1000 shards.

split.inverse-sampling.rate

The inverse of the sampling rate used in the sample sharding strategy. For example, if this value is set to 1000, it means a 1/1000 sampling rate is applied during the sampling process. This option provides flexibility in controlling the granularity of the sampling, thus affecting the final number of shards. It’s especially useful when dealing with very large datasets where a lower sampling rate is preferred. The default value is 1000.

partition_column [string]

The column name for split data.

partition_upper_bound [BigDecimal]

The partition_column max value for scan, if not set SeaTunnel will query database get max value.

partition_lower_bound [BigDecimal]

The partition_column min value for scan, if not set SeaTunnel will query database get min value.

partition_num [int]

Not recommended for use, The correct approach is to control the number of split through split.size

How many splits do we need to split into, only support positive integer. default value is job parallelism.

tips

If the table can not be split(for example, table have no Primary Key or Unique Index, and partition_column is not set), it will run in single concurrency.

Use table_path to replace query for single table reading. If you need to read multiple tables, use table_list.

appendix

there are some reference value for params above.

Example

simple

Jdbc {
    url = "jdbc:mysql://localhost/test?serverTimezone=GMT%2b8"
    driver = "com.mysql.cj.jdbc.Driver"
    connection_check_timeout_sec = 100
    user = "root"
    password = "123456"
    query = "select * from type_bin"
}

parallel by partition_column

env {
  parallelism = 10
  job.mode = "BATCH"
}
source {
    Jdbc {
        url = "jdbc:mysql://localhost/test?serverTimezone=GMT%2b8"
        driver = "com.mysql.cj.jdbc.Driver"
        connection_check_timeout_sec = 100
        user = "root"
        password = "123456"
        query = "select * from type_bin"
        partition_column = "id"
        split.size = 10000
        # Read start boundary
        #partition_lower_bound = ...
        # Read end boundary
        #partition_upper_bound = ...
    }
}
sink {
  Console {}
}

Parallel Boundary:

It is more efficient to specify the data within the upper and lower bounds of the query. It is more efficient to read your data source according to the upper and lower boundaries you configured.

source {
    Jdbc {
        url = "jdbc:mysql://localhost:3306/test?serverTimezone=GMT%2b8&useUnicode=true&characterEncoding=UTF-8&rewriteBatchedStatements=true"
        driver = "com.mysql.cj.jdbc.Driver"
        connection_check_timeout_sec = 100
        user = "root"
        password = "123456"
        # Define query logic as required
        query = "select * from type_bin"
        partition_column = "id"
        # Read start boundary
        partition_lower_bound = 1
        # Read end boundary
        partition_upper_bound = 500
        partition_num = 10
        properties {
         useSSL=false
        }
    }
}

parallel by Primary Key or Unique Index

Configuring table_path will turn on auto split, you can configure split.* to adjust the split strategy

env {
  parallelism = 10
  job.mode = "BATCH"
}
source {
    Jdbc {
        url = "jdbc:mysql://localhost/test?serverTimezone=GMT%2b8"
        driver = "com.mysql.cj.jdbc.Driver"
        connection_check_timeout_sec = 100
        user = "root"
        password = "123456"
        table_path = "testdb.table1"
        query = "select * from testdb.table1"
        split.size = 10000
    }
}
sink {
  Console {}
}

multiple table read:

Configuring table_list will turn on auto split, you can configure `split.` to adjust the split strategy*

Jdbc {
    url = "jdbc:mysql://localhost/test?serverTimezone=GMT%2b8"
    driver = "com.mysql.cj.jdbc.Driver"
    connection_check_timeout_sec = 100
    user = "root"
    password = "123456"
    table_list = [
        {
          # e.g. table_path = "testdb.table1"、table_path = "test_schema.table1"、table_path = "testdb.test_schema.table1"
          table_path = "testdb.table1"
        },
        {
          table_path = "testdb.table2"
          # Use query filetr rows & columns
          query = "select id, name from testdb.table2 where id > 100"
        }
    ]
    #where_condition= "where id > 100"
    #split.size = 10000
    #split.even-distribution.factor.upper-bound = 100
    #split.even-distribution.factor.lower-bound = 0.05
    #split.sample-sharding.threshold = 1000
    #split.inverse-sampling.rate = 1000
}

Changelog

2.2.0-beta 2022-09-26

• Add ClickHouse Source Connector

2.3.0-beta 2022-10-20

• [Feature] Support Phoenix JDBC Source (2499)
• [Feature] Support SQL Server JDBC Source (2646)
• [Feature] Support Oracle JDBC Source (2550)
• [Feature] Support StarRocks JDBC Source (3060)
• [Feature] Support GBase8a JDBC Source (3026)
• [Feature] Support DB2 JDBC Source (2410)

next version

• [BugFix] Fix jdbc split bug (3220)
• [Feature] Support Sqlite JDBC Source (3089)
• [Feature] Support Tablestore Source (3309)
• [Feature] Support Teradata JDBC　Source (3362)
• [Feature] Support JDBC Fetch Size Config (3478)
• [Feature] Support Doris JDBC Source (3586)
• [Feature] Support Redshift JDBC Sink(#3615)
• [BugFix] Fix jdbc connection reset bug (3670)
• [Improve] Add Vertica connector(#4303)