PostgreSQL Migration and Synchronization to ClickHouse

NineData data replication supports copying data from self-built PostgreSQL to ClickHouse, supporting both structure and full data replication.

Feature Introduction

NineData data replication supports high-performance replication of structure, full data, and incremental data between data sources. For MySQL data sources, it also provides bidirectional replication, enabling the quick construction of geo-distributed active-active business architectures.

• Structure: Supports the replication of object structures between homogeneous and heterogeneous data sources, greatly reducing the barrier to data replication between two sources.
• Full Data: Achieves row-level concurrent batch replication through intelligent data sharding, effectively ensuring replication performance. The independently developed novel breakpoint resume technology ensures the accuracy of data in tables without primary keys.
• Incremental Data: Supports incremental data replication of full object types DML|DDL, combined with transaction-level concurrency, hotspot merging, and other technologies to provide robust replication performance while ensuring transaction consistency.
• Bidirectional Real-time Data Replication (only between MySQL instances): Direct bidirectional replication of data between multiple nodes ensures that all node data remains up-to-date.

With the above features, it is easy and efficient to achieve scenarios such as full|incremental data replication, full|incremental data migration, full|incremental data synchronization, data integration, and seamless migration without downtime, providing enterprises with flexible and reliable data replication solutions.

Prerequisites

• Both source and target data sources must be added to NineData. For instructions on how to add data sources, please refer to Adding Data Sources.

• The versions of the source and target data sources are as follows:

  Source Data Source	Target Data Source
  PostgreSQL 9 or above	ClickHouse 20.8 or above

• For incremental replication, open the postgresql.conf file and configure the following parameters. If you cannot find the location of this file, you can execute the SHOW config_file; SQL command in the psql client to view it.

  • The wal_level parameter of the source data source must be logical. To confirm the current value, execute SHOW wal_level directly in the client.
  • Set the wal_sender_timeout parameter of the source data source to 0. This parameter is used to interrupt replication connections that have been stagnant for more than the specified number of milliseconds. The default value is 60000 milliseconds. Setting it to 0 will disable the timeout mechanism. To confirm the current value, execute SHOW wal_sender_timeout directly in the client.
  • The max_replication_slots parameter of the source data source must be greater than 1. This parameter specifies the maximum number of replication slots that the server can support. The default value is 10.
  • The max_wal_senders parameter of the source data source must be greater than 1. This parameter specifies the maximum number of concurrent connections. The default value is 10. To confirm the current value, execute SHOW max_wal_senders directly in the client.

• The access_management parameter value for the account used to log in to ClickHouse must be 1.

  Configuration Method:

  Open the User.xml configuration file located in the /etc/clickhouse-server directory with an editor, find the target user, and add the parameter: <access_management>1</access_management>.

• The following permissions are required for both the source and target data sources.

  Replication Type	Source Data Source Permissions	Target Data Source Permissions
  Full replication	CONNECT, SELECT	Permissions related to TABLE (CREATE, CREATE TABLE, ALTER, ALTER COLUMN, DROP, SELECT, INSERT)

Considerations

• If there are unsigned integer type fields in the source PostgreSQL data source, they will be mapped to Uint type in ClickHouse.

• Typically, the replication scenario from PostgreSQL to ClickHouse needs to include structural replication, where the system will automatically migrate the table structure from PostgreSQL to ClickHouse and insert two system columns into the replicated tables in ClickHouse. If the replication type does not include structural replication, you need to manually ensure the following:

  • The table structures in ClickHouse data sources involved in replication must be consistent with those in PostgreSQL data sources.

  • Two system columns are manually inserted into the replicated tables in ClickHouse to record DML operations and the time of Binlog.

    Column Name	Data Type	Default Value	Description
    _jz_data_sign	Int8	DEFAULT 1	Records the type of DML operation to ensure consistency between data in ClickHouse and PostgreSQL. INSERT operation: Record as 1. DELETE operation: Record as -1. UPDATE operation: Split into INSERT and DELETE two records.
    _jz_data_time	String	DEFAULT now()	Records the time of Binlog.

    Example:

    lessCopy code
    CREATE TABLE [IF NOT EXISTS] [db.]table_name [ON CLUSTER cluster]
    (
        ...
        _jz_data_sign Int8 DEFAULT 1 COMMENT 'replication data update sign: add = 1, remove = -1',
        _jz_data_time String DEFAULT now() COMMENT 'replication data update time'
    ) ENGINE = engine

Usage Restrictions

• Evaluate the performance of both the source and target data sources before executing data synchronization. It is recommended to perform data synchronization during low traffic periods to avoid increasing the database load during full data initialization.
• It is recommended to ensure that each table in the synchronization object has a primary key or unique constraint, and column names are unique to prevent duplicate synchronization of the same data.

Operation Steps

1. Log in to the NineData console.

2. Click on in the left navigation pane.

3. On the page, click on the button in the upper right corner.

4. On the tab, configure according to the table below, and click .

   Parameter	Description
   	Enter the name of the data synchronization task. For easy retrieval and management later, please use a meaningful name. Up to 64 characters are supported.
   	The data source where the synchronization object resides.
   	The data source that receives the synchronized object.
   	Choose the content to be copied to the target data source. : Only synchronize the database table structure of the source data source, without synchronizing data. : Synchronize all objects and data of the source data source, i.e., full data replication. The switch on the right side is for periodic full replication. For more information, please refer to Periodic Full Replication.
   (Select when is selected)	: Stop the task when a table with the same name is detected during the pre-check stage. : When a table with the same name is detected during the pre-check stage, a prompt is sent, and the task continues. During structural replication, ignore this table with the same name. If you are also performing data replication, the data will be appended to the table with the same name without overwriting the existing data. : When a table with the same name is detected during the pre-check stage, a prompt is sent, and the task continues. During structural replication, the table with the same name in the target database is deleted, and the table structure is re-copied based on the source database. If you are also performing data replication, the data will be written after the table structure replication is completed. (Optional when performing both structural and data replication): When a table with the same name is detected during the pre-check stage, a prompt is sent, and the task continues. During structural replication, the table structure in the target database is retained, and the data in the table with the same name is cleared when data replication starts, then copied again from the original table.
   (Select when is not selected)	: Stop the task when data exists in the target table during the pre-check stage. : When data exists in the target table during the pre-check stage, ignore this part of the data and append other data. : When data exists in the target table during the pre-check stage, delete this part of the data and rewrite it.

5. On the tab, configure the following parameters, then click .

   Parameter	Description
   	Choose the content to be copied. You can choose to copy all contents of the source database, or . In the list, select the content you want to copy, and click > to add it to the list on the right.

6. On the tab, select different operations based on the selected replication type, then click . If there are updates in the source and target data sources during the configuration mapping stage, you can click the button in the upper right corner of the page to retrieve the information of the source and target data sources again.

   • Includes : Configure the table name after synchronization to the target data source.

     tip

     You can click on the right side of the page to customize the name of the column after synchronization to the target data source. In addition, you can set to configure filtering conditions through comparison expressions, and only data that meets the filtering conditions will be synchronized to the target data source. For example, if you set the filtering condition to emp_no>=10005, data with emp_no less than 10005 will not be synchronized to the target data source.

   • Excludes : The system automatically selects the same-name database in the target data source. If it does not exist, you need to manually select the target database. The table names and column names in the target database need to be consistent with the synchronization objects. If they are not consistent, you can manually map the table names and column names.

7. On the tab, wait for the system to complete the pre-check. After the pre-check passes, click .

   tip

   • You can check . After the synchronization task is completed, automatically start data consistency comparison based on the source data source to ensure consistency of data on both ends. Depending on your selection of , the timing of starting is as follows:
     • : Start after structural replication is completed.
     • + , : Start after full replication is completed.
   • If the pre-check fails, you need to click on the right side of the target check item in the column, troubleshoot the reason for the failure, manually fix it, then click to re-perform the pre-check until it passes.
   • For check items with as , you can repair or ignore them according to the specific situation.

8. On the page, when prompted with , the synchronization task starts running. At this point, you can:

   • Click to view the execution status of each stage of the synchronization task.
   • Click to return to the task list page.

Viewing Synchronization Results

1. Log in to the NineData console.

2. Click on > in the left navigation pane.

3. On the page, click on the of the target synchronization task, and the page description is as follows.

   

   Number	Function	Description
   1	Configure Alerts	After configuring alerts, the system will notify you in the selected way when the task fails. For more information, please refer to Operational Monitoring Overview.
   2	More	: Pause the task. Only tasks with the status Running are selectable. : Create a new replication task with the same configuration as the current task. : End tasks that are incomplete or in listening (i.e., in incremental synchronization). After terminating the task, it cannot be restarted, so please proceed with caution. If triggers are included in the synchronization object, trigger replication options will pop up, please choose as needed. : Delete the task. Once the task is deleted, it cannot be recovered, so please proceed with caution.
   3	Structural Replication (Displayed in scenarios involving structural replication)	Display the progress and details of structural replication. Click on on the right side of the page: View the execution log of structural replication. Click on on the right side of the page: View the latest information. Click on in the column on the right side of the target object in the list: View SQL playback.
   4	Full Replication (Displayed in scenarios involving full replication)	Display the progress and details of full replication. Click on on the right side of the page: View various monitoring indicators during full replication. During full replication, you can also click on on the right side of the monitoring indicator page to limit the rate of data written to the target data source per second. The unit is rows/second. Click on on the right side of the page: View the execution log of full replication. Click on on the right side of the page: View the latest information.
   5	Data Comparison	Display the comparison results between the source data source and the target data source. If you have not enabled data comparison, click on the page. Click on on the right side of the page: Re-initiate the comparison of data between the current source and target ends. Click on on the right side of the page: After starting the comparison task, you can click this button to stop the comparison task immediately. Click on on the right side of the page: View the execution log of consistency comparison. Click on (only displayed in data comparison): View the trend chart of RPS (records per second compared) for comparison. Click on to view records from earlier times. Click on in the column on the right side of the comparison list (displayed only under the tab when there is inconsistency): View detailed comparison between the source and target ends. Click on in the column on the right side of the comparison list (displayed only in case of inconsistency): Generate change SQL, which you can directly copy to the target data source to execute and modify the inconsistent content.
   6	Expand	Display detailed information of the current replication task.

Appendix: Checklist Overview

Check Item	Check Content
Source datasource connection check	Check the status of the gateway of the source datasource, database connectable, and verify the username and password
Target datasource connection check	Check the status of the gateway of the target datasource, database connectable, and verify the username and password
Target databse privilege check	Check whether the account privileges of the target database meet the requirements
Source databse privilege check	Check whether the account privileges of the source database meet the requirements
Target database data existence check	Check whether the object to be replicated not empty in the target database
Objects with the same name in the target database existence check	Check whether the object to be replicated already exists in the target database

Related Documents

Introduction to Data Replication