Updated on 2024-04-11 GMT+08:00

From PostgreSQL to PostgreSQL

Supported Source and Destination Databases

Table 1 Supported databases

Source DB

Destination DB

  • On-premises database (PostgreSQL 9.4, 9.5, 9.6, 10, 11, 12, 13 and 14)
  • ECS database (PostgreSQL 9.4, 9.5, 9.6, 10, 11, 12, 13 and 14)
  • Other cloud database (PostgreSQL 9.4, 9.5, 9.6, 10, 11, 12, 13 and 14)
  • RDS for PostgreSQL (9.5, 9.6, 10, 11, 12, 13, 14)

RDS for PostgreSQL (9.5, 9.6, 10, 11, 12, 13, 14)

NOTE:

The major version of the destination database must be the same as or later than that of the source database.

Supported Synchronization Objects

Table 2 lists the objects that can be synchronized in different scenarios. DRS will automatically check the objects you selected before the synchronization.

Table 2 Supported synchronization objects

Type

Notes

Objects

  • Instance-level synchronization is not supported. Only one database can be synchronized at a time. Multiple DRS tasks are required to synchronize multiple databases.
  • Supported field types:

    Digit, currency, character, binary, date/time, boolean, enumeration, geometry, network address, bit, text search, UUID, XML, JSON, array, compound, and range.

  • Scope of full synchronization
    • The following objects are supported in the database-level synchronization: schemas, tables, indexes, constraints, views, materialized views, sequences, stored procedures, rules, triggers, foreign keys, sorting rules, plug-ins, code conversion information, aggregate functions, operators, statistics extension, conversion information, text search configurations, functions, data types, type conversion, users, event triggers, text search parsers, and text search templates

      During the table-level synchronization, only tables, views, materialized views, sequences, users, and common indexes can be synchronized.

      During object file import, tables can be synchronized.

    • Objects that are not supported: system schemas (schemas starting with pg_, information_schema, sys, utl_raw, dbms_lob, dbms_output, and dbms_random), system catalogs, system users, tablespaces, foreign-data wrappers, foreign servers, user mappings, publications, and subscriptions
    NOTE:

    The restrictions on the objects that can be synchronized are as follows:

    • Object name: The database name cannot contain +" %?\<>, the schema name and table name cannot contain ".'<>, and the column name cannot contain double quotation marks (") and single quotation marks (').
    • Table: Temporary tables are not synchronized. During table-level synchronization, table constraints, indexes, and rules are synchronized, except for table triggers.
    • Schema: Permissions of the public schema are not synchronized. During table-level synchronization, the permissions of existing schemas in the destination database are synchronized.
    • Function: Do not synchronize C language functions or functions with the leakproof or support attribute.
    • Plug-in: The metadata of plug-ins is not synchronized.
    • Data type: Basic data types are not synchronized.
    • Type conversion: The binary coercion type cannot be converted.
    • Event trigger: Event triggers can be synchronized only when the destination database version is RDS for PostgreSQL 11.11 or later.
    • Text search parser: Text search parsers can be synchronized only when the destination database version is RDS for PostgreSQL 11.11 or later.
    • Text search template: Text search templates can be synchronized only when the destination database version is RDS for PostgreSQL 11.11 or later.
    • User: Existing users in the destination database, superuser, replication, and bypassrls attributes of users, and member relationships of superuser users are not synchronized. If the object owner or grantor is superuser, its owner or grantor is not synchronized. If the destination database is RDS for PostgreSQL DB instance, the password of the user to be synchronized cannot contain the username. During table-level synchronization, the default access permissions of source database users are not synchronized.
  • Scope of incremental synchronization
    • Some DML statements, including INSERT, UPDATE, and DELETE, can be synchronized.
    • Some DDL statements can be synchronized, including TRUNCATE (only for PostgreSQL 11 or later), CREATE SCHEMA, CREATE TABLE, DROP TABLE, ALTER TABLE (including ADD COLUMN, DROP COLUMN, ALTER COLUMN, RENAME COLUMN, ADD CONSTRAINT, DROP CONSTRAINT and RENAME), CREATE SEQUENCE, DROP SEQUENCE, ALTER SEQUENCE, CREATE INDEX, ALTER INDEX, DROP INDEX, CREATE VIEW, ALTER VIEW, COMMENT ON COLUMN, COMMENT ON TABLE, COMMENT ON SCHEMA, COMMENT ON SEQUENCE, COMMENT ON INDEX, and COMMENT ON VIEW.

      During table-level synchronization, only the following DDL operations can be synchronized: TRUNCATE (only for PostgreSQL 11 or later), DROP TABLE, COMMENT ON COLUMN, COMMENT ON TABLE, and ALTER TABLE (including ADD COLUMN, DROP COLUMN, ALTER COLUMN, RENAME COLUMN, ADD CONSTRAINT, DROP CONSTRAINT and RENAME).

    • Not synchronized: DML statements of unlogged tables and temporary tables
      NOTE:

      The source database captures DDL statements using event triggers and records them in specific tables, so you need to create event triggers and functions in the source database in advance. For details, see Creating Triggers and Functions to Implement Incremental DDL Synchronization for PostgreSQL.

Database Account Permission Requirements

To start a synchronization task, the source and destination database users must meet the requirements in the following table. Different types of synchronization tasks require different permissions. For details, see Table 3. DRS automatically checks the database account permissions in the pre-check phase and provides handling suggestions.

  • You are advised to create an independent database account for DRS task connection to prevent task failures caused by database account password modification.
  • After changing the account passwords for the source or destination databases, modify the connection information in the DRS task as soon as possible to prevent automatic retry after a task failure. Automatic retry will lock the database accounts.
Table 3 Database account permission

Type

Full

Full+Incremental

Source database user

The CONNECT permission for databases, the USAGE permission for schemas, the SELECT permission for tables, the SELECT permission for sequences, and the SELECT permission for system table catalog pg_catalog.pg_authid (used for synchronizing user passwords)

The CONNECT permission for databases, the USAGE permission for schemas, the SELECT permission for tables, the SELECT permission for sequences, the SELECT permission for system table catalog pg_catalog.pg_authid (used for synchronizing user passwords), the UPDATE, DELETE, and TRUNCATE permissions for tables that do not have primary keys, and the permission to create replication connections

NOTE:
  • The UPDATE, DELETE, and TRUNCATE permissions for tables that do not have primary keys are only used to temporarily lock tables to ensure data consistency after the migration.
  • To add the permission to create replication connections, perform the following steps:
    1. Add host replication <src_user_name> <drs_instance_ip>/32 md5 before all configurations in the pg_hba.conf file of the source database.
    2. Run select pg_reload_conf(); in the source database as user SUPERUSER, or restart the DB instance to apply the changes.

Destination database user

  • Database-level:
    • If the destination database is not PostgreSQL, the CREATEDB permission is required.
    • If the destination database is PostgreSQL, the CONNECT and CREATE permissions on PostgreSQL databases and the USAGE and CREATE permissions on public schemas are required.
  • Table-level:
    • To synchronize databases, the CREATEDB permission is required.
    • To synchronize a schema, the CONNECT and CREATE permissions for the database that contains the schema are required.
    • To synchronize objects in a schema, the CONNECT permission for the database that contains the schema, and the USAGE and CREATE permissions for the schema that contain the object are required.
  • Synchronization user: The CREATEROLE permission is required.
  • Synchronization user permissions: The default privilege cannot be modified. Otherwise, the object permissions of the destination database may be inconsistent with those of the source database.
NOTE:

To synchronize event triggers, text search parsers, and text search templates, the destination database version must be RDS for PostgreSQL 11.11 or later, and the destination database user must be user root or a member of user root.

Suggestions

  • When a task is being started or in the full synchronization phase, do not perform DDL operations on the source database. Otherwise, the task may be abnormal.
  • To keep data consistency before and after the synchronization, ensure that no data is written to the destination database during the synchronization.
  • The success of database synchronization depends on environment and manual operations. To ensure a smooth synchronization, perform a synchronization trial before you start the synchronization to help you detect and resolve problems in advance.
  • Start your synchronization task during off-peak hours. A less active database is easier to synchronize successfully. If the data is fairly static, there is less likely to be any severe performance impacts during the synchronization.
    • If network bandwidth is not limited, the query rate of the source database increases by about 50 MB/s during full synchronization, and two to four CPUs are occupied.
    • To ensure data consistency, tables to be synchronized without a primary key may be locked for 3s.
    • The data being synchronized may be locked by other transactions for a long period of time, resulting in read timeout.
    • When DRS concurrently reads data from a database, it will use about 6 to 10 sessions. The impact of the connections on services must be considered.
    • If you read a table, especially a large table, during the full migration, the exclusive lock on that table may be blocked.
  • Data-Level Comparison

    To obtain accurate comparison results, start data comparison at a specified time point during off-peak hours. If it is needed, select Start at a specified time for Comparison Time. Due to slight time difference and continuous operations on data, data inconsistency may occur, reducing the reliability and validity of the comparison results.

Precautions

The full+incremental synchronization process consists of four phases: task startup, full synchronization, incremental synchronization, and task completion. (A single incremental synchronization task or a single full synchronization task contains three phases.) To ensure smooth synchronization, read the following notes before creating a synchronization task.

Table 4 Precautions

Type

Constraints

Starting a task

  • Source database parameter requirements:
    • The partition table trigger of the source database cannot be set to disable.
    • For a full synchronization task, the source database can be a standby database, but hot_standby_feedback must be set to on. For an incremental synchronization task, the source database cannot be a standby database.
    • To perform incremental synchronization:
      The pg_hba.conf file of the source database contains the following configuration:
      host replication all 0.0.0.0/0 md5

      The wal_level value of the source database must be logical.

      The test_decoding plug-in has been installed on the source database.

      The replica identity attribute of tables that do not have primary keys in the source database must be full.

      The max_replication_slots value of the source database must be greater than the number of used replication slots.

      The max_wal_senders value of the source database must be greater than or equal to the max_replication_slots value.

      If the toast attribute of the primary key column in the source database is main, external, or extended, the replica identity attribute must be full.

  • Source database object requirements:
    • Triggers with the same name cannot exist in the source database.
    • The objects that have dependencies must be synchronized at the same time. Otherwise, the synchronization may fail.
  • Destination database parameter requirements:
    • The block_size value of the destination database must be greater than or equal to that of the source database.
    • The lc_monetary values of the source and destination databases must be the same.
    • To perform incremental synchronization, if the synchronization object contains foreign keys, triggers, or event triggers, set session_replication_role of the destination database to replica. After the synchronization is complete, change the value to origin.
  • Destination database object requirements:
    • The destination database cannot contain objects with the same type and name as the objects to be synchronized, including databases, schemas, and tables. System databases, system schemas, and system tables are excluded.
    • The destination table can contain more columns than the source table. However, the following failures must be avoided:

      Assume that extra columns on the destination cannot be null or have default values. If newly inserted data records are synchronized from the source to the destination, the extra columns will become null, which does not meet the requirements of the destination and will cause the task to fail.

      Assume that extra columns on the destination must be fixed at a default value and have a unique constraint. If newly inserted data records are synchronized from the source to the destination, the extra columns will contain multiple default values. That does not meet the unique constraint of the destination and will cause the task to fail.

  • Other notes:
    • When a schema name or table name is mapped, to prevent conflicts between indexes and constraint names, the original index name in the table is changed to the following format after synchronization: i_+hash value+original index name (which may be truncated)+_key The hash value is calculated based on the original schema name_original table name_original index name. Similarly, the original constraint name on the table is changed to c_ + hash value + original constraint name (which may be truncated) + _key.
    • Before starting a full+incremental or incremental synchronization task, ensure that no long transaction is started in the source database. Starting the long transaction will block the creation of the logical replication slot and cause the task to fail.
    • For a full+incremental or incremental synchronization task, if an internal error occurs during the pre-check and the task stops before it is started, check and delete the streaming replication slot by referring to Forcibly Stopping Synchronization of PostgreSQL to avoid residual streaming replication slots in the source database.
    • If you choose to synchronize DDL statements, ensure that the DDL statements executed on the source database are compatible with the destination database.
      NOTE:

      DDL statements are captured using event triggers in the source database, recorded in a specific table, and then synchronized to the destination database. You need to create event triggers and functions in the source database before starting a task. For details, see Creating Triggers and Functions to Implement Incremental DDL Synchronization for PostgreSQL.

Full synchronization

  • Do not change the port of the source and destination databases, or change or delete the passwords and permissions of the source and destination database users. Otherwise, the task may fail.
  • Do not run any DDL statement in the source database. Otherwise, data may be inconsistent or the task may fail.
  • Do not write data to the destination database. Otherwise, data may be inconsistent.

Incremental synchronization

  • Do not change the port of the source and destination databases, or change or delete the passwords and permissions of the source and destination database users. Otherwise, the task may fail.
  • Do not change the primary key or unique key (if the primary key does not exist) of the source database table. Otherwise, incremental data may be inconsistent or the task may fail.
  • Do not modify the replica identity attribute of tables in the source database. Otherwise, incremental data may be inconsistent or the task may fail.
  • Do not write data to the destination database. Otherwise, data may be inconsistent.
  • During database-level synchronization, if a table without a primary key is added to the source database, you must set replica identity of the table to full before writing data. Otherwise, data may be inconsistent or the task may fail.
  • During database-level synchronization, if a primary key table is added to the source database and the toast attribute of the primary key column is main, external, or extended, the replica identity attribute of the table must be set to full before writing data. Otherwise, data may be inconsistent or the task may fail.

Synchronization comparison

  • You are advised to compare data in the source database during off-peak hours to prevent inconsistent data from being falsely reported and reduce the impact on the source database and DRS tasks.
  • During incremental synchronization, if data is written to the source database, the comparison results may be inconsistent.
  • Data cannot be compared during full synchronization.
  • Do not limit the synchronization speed during data comparison.

Stopping a task

  • Stop a task normally:
    • The destination database sequence value is automatically reset. The auto-increment sequence value is the source database sequence value plus the security margin, and the auto-decrement sequence value is the source database sequence value minus the security margin. The default security margin is 10,000. If users are synchronized, the user memberships are automatically synchronized after the task is complete.
    • When a full+incremental synchronization task is complete, the streaming replication slot created by the task in the source database is automatically deleted.
    • If the value of destination database session_replication_role is replica when the full+incremental synchronization task is complete, change the value to origin.
  • Forcibly stop a task:
    • You need to manually update the sequence value in the destination database. For details, see Forcibly Stopping Synchronization of PostgreSQL.
    • To forcibly stop a full+incremental real-time synchronization task, you need to manually delete the replication slots that may remain in the source database. For details, see Forcibly Stopping Synchronization of PostgreSQL.
    • If the value of destination database session_replication_role is replica, change it to origin to forcibly stop the full+incremental synchronization task.
    • The naming rule of a logic replication slot is drs_unique_ID. To obtain the unique ID, replace the hyphen (-) in the task node ID with an underscore (_). You can find the node ID in the task node id is xxxx log on the Synchronization Logs page.

Prerequisites

  • You have logged in to the DRS console.
  • For details about the DB types and versions supported by real-time synchronization, see Real-Time Synchronization.

Procedure

This section uses to-the-cloud synchronization from PostgreSQL to PostgreSQL as an example to describe how to configure a real-time synchronization task in the VPC network scenario.

  1. On the Data Synchronization Management page, click Create Synchronization Task.
  2. On the Create Synchronization Instance page, specify the task name, description, and the synchronization instance details, and click Create Now.

    • Task information description
      Table 5 Task and recipient description

      Parameter

      Description

      Task Name

      The task name must start with a letter and consist of 4 to 50 characters. It can contain only letters, digits, hyphens (-), and underscores (_).

      Description

      The description consists of a maximum of 256 characters and cannot contain special characters !=<>'&"\

    • Synchronization instance details
      Table 6 Synchronization instance settings

      Parameter

      Description

      Data Flow

      Select To the cloud.

      Source DB Engine

      Select PostgreSQL.

      Destination DB Engine

      Select PostgreSQL.

      Network Type

      Available options: VPC, Public network and VPN or Direct Connect. VPC is used as an example.

      • VPC is suitable for data synchronization between cloud databases of the same account in the same region.
      • Public network is suitable for data synchronization from on-premises or external cloud databases to the destination databases bound with an EIP.
      • VPN or Direct Connect is suitable for synchronization of data between on-premises databases and cloud databases, between cloud databases of different accounts in the same region, or between cloud databases across regions.

      Destination DB Instance

      The RDS for PostgreSQL DB instance.

      Synchronization Instance Subnet

      Select the subnet where the synchronization instance is located. You can also click View Subnet to go to the network console to view the subnet where the instance resides.

      By default, the DRS instance and the destination DB instance are in the same subnet. You need to select the subnet where the DRS instance resides and ensure that there are available IP addresses. To ensure that the synchronization instance is successfully created, only subnets with DHCP enabled are displayed.

      Synchronization Mode

      • Full+Incremental

        This synchronization mode allows you to synchronize data in real time. After a full synchronization initializes the destination database, an incremental synchronization parses logs to ensure data consistency between the source and destination databases.

      • Full

        All objects and data in non-system databases are synchronized to the destination database at a time. This mode is applicable to scenarios where service interruption is acceptable.

    If a task fails to be created, DRS retains the task for three days by default. After three days, the task automatically ends.

  3. On the Configure Source and Destination Databases page, wait until the synchronization instance is created. Then, specify source and destination database information and click Test Connection for both the source and destination databases to check whether they have been connected to the synchronization instance. After the connection tests are successful, select the check box before the agreement and click Next.

    Establish the connectivity between the DRS instance and the source and destination databases.

    • Network connectivity: Ensure that the source and destination databases accept connections from the IP address of the DRS instance. To access databases over a public network, configure the database to accept connections from the EIP of the DRS instance. To access databases over a VPC, VPN, or Direct Connect network, configure the database to accept connections from the private IP address of the DRS instance.
    • Account connectivity: Ensure that the source and destination databases allows connections from the DRS instance using the username and password.

    The source database can be an ECS database or an RDS instance. Configure parameters based on the database type.

    • Scenario 1: Databases on an ECS - source database configuration
      Table 7 Self-build on ECS - source database information

      Parameter

      Description

      Source Database Type

      Select Self-built on ECS.

      VPC

      A dedicated virtual network in which the source database is located. It isolates networks for different services. You can select an existing VPC or create a VPC.

      Subnet

      A subnet provides dedicated network resources that are isolated from other networks, improving network security. The subnet must be in the AZ where the source database resides. You need to enable DHCP for creating the source database subnet.

      IP Address or Domain Name

      The IP address or domain name of the source database.

      Port

      The port of the source database. Range: 1 – 65535

      Database Name

      Indicates whether to specify a database. If this option is enabled, enter the database name.

      Database Username

      The username for accessing the source database.

      Database Password

      The password for the database username.

      SSL Connection

      SSL encrypts the connections between the source and destination databases.

      The IP address, domain name, username, and password of the source database are encrypted and stored in DRS, and will be cleared after the task is deleted.

    • Scenario 2: RDS DB instance - source database configuration
      Table 8 RDS DB instance - source database information

      Parameter

      Description

      Source Database Type

      Select an RDS DB instance.

      DB Instance Name

      Select the RDS PostgreSQL instance to be synchronized as the source DB instance.

      Database Username

      The username for accessing the source database.

      Database Password

      The password for the database username.

    Table 9 Destination database settings

    Parameter

    Description

    DB Instance Name

    The RDS PostgreSQL instance you selected when creating the migration task and cannot be changed.

    Database Username

    The username for accessing the destination database.

    Database Password

    The password for the database username.

    The username and password of the source and destination databases are encrypted and stored in the databases and the synchronization instance during the synchronization. After the task is deleted, the username and password are permanently deleted.

  4. On the Set Synchronization Task page, select the synchronization objects and accounts and click Next.

    Table 10 Synchronization Object

    Parameter

    Description

    Flow Control

    You can choose whether to control the flow.

    • Yes

      You can customize the maximum synchronization speed.

      In addition, you can set the time range based on your service requirements. The traffic rate setting usually includes setting of a rate limiting time period and a traffic rate value. Flow can be controlled all day or during specific time ranges. The default value is All day. A maximum of three time ranges can be set, and they cannot overlap.

      The flow rate must be set based on the service scenario and cannot exceed 9,999 MB/s.

    • No
      The synchronization speed is not limited and the outbound bandwidth of the source database is maximally used, which will increase the read burden on the source database. For example, if the outbound bandwidth of the source database is 100 MB/s and 80% bandwidth is used, the I/O consumption on the source database is 80 MB/s.
      NOTE:
      • The flow control mode takes effect only in the full synchronization phase.
      • You can also change the flow control mode after creating a task. For details, see Modifying the Flow Control Mode.

    Incremental Conflict Policy

    The conflict policy refers to the conflict handling policy during incremental synchronization. By default, conflicts in the full synchronization phase are ignored. Select any of the following conflict policies:

    • Ignore

      The system will skip the conflicting data and continue the subsequent synchronization process.

    • Report error

      The synchronization task will be stopped and fail.

    • Overwrite

      Conflicting data will be overwritten.

    Synchronize

    Available options: Index, Incremental DDLs, and Populate materialized views during the full synchronization phase

    Populate materialized views during the full synchronization phase: This option takes effect only for materialized views that was populated in the source database. This operation affects the full synchronization performance. You perform this operation after the full synchronization is complete.

    Synchronization Object

    The left pane displays the source database objects, and the right pane displays the selected objects. You can select Tables, Import object file, or Databases for Synchronization Object as required.

    • Database-level synchronization: In full synchronization, the selected databases and the inventory data of the database objects are synchronized. In incremental synchronization, the DML and some DDL statements of all tables except unlogged tables and temporary tables are synchronized.
    • Table-level synchronization: In full synchronization, the inventory data of the selected tables, sequences, views, or materialized views is synchronized. In incremental synchronization, the DML and some DDL statements of the selected tables are synchronized.
    • If the synchronization objects in source and destination databases have different names, you can map the source object name to the destination one. For details, see Mapping Object Names.

      When a schema name or table name is mapped, to prevent conflicts between indexes and constraint names, the original index name in the table is changed to the following format after synchronization: i_+hash value+original index name (which may be truncated)+_key The hash value is calculated based on the original schema name_original table name_original index name. Similarly, after the synchronization, the original constraint name on the table is changed to c_ + hash value + original constraint name (which may be truncated) + _key.

    • For details about how to import an object file, see Importing Synchronization Objects.
    NOTE:
    • To quickly select the desired database objects, you can use the search function.
    • If there are changes made to the source databases or objects, click in the upper right corner to update the objects to be synchronized.
    • If the object name contains spaces, the spaces before and after the object name are not displayed. If there are multiple spaces between the object name and the object name, only one space is displayed.
    • The name of the selected synchronization object cannot contain spaces.

    Synchronize Account

    During the synchronization, you can synchronize accounts based on your service requirements. For details, see Table 11.

    Table 11 Accounts and permissions to be synchronized

    Parameter

    Description

    Account

    Account name of the source database.

    Whether to Support Synchronization

    Whether the account can be synchronized.

    There are accounts that can be synchronized and accounts that cannot be synchronized. For an account that cannot be synchronized, the specific reason is displayed in View in the Remarks column.

    Parent Account

    Parent account.

    Parent Account That Cannot Be Synchronized

    The parent account that cannot be synchronized.

    Account Attribute

    Attributes of the source database account.

    Account Attribute That Cannot Be Synchronized

    The account attributes that cannot be synchronized due to insufficient permissions of the destination database user.

    Remarks

    Description of the parent account and account attributes that cannot be synchronized. You can go to the next step only after confirming all remarks.

    Synchronize object permissions

    Whether to synchronize permissions corresponding to the account.

  5. On the Check Task page, check the synchronization task.

    • If any check fails, review the cause and rectify the fault. After the fault is rectified, click Check Again.
    • If all check items are successful, click Next.

      You can proceed to the next step only when all checks are successful. If there are any items that require confirmation, view and confirm the details first before proceeding to the next step.

  6. On the displayed page, specify Start Time, confirm that the configured information is correct, and click Submit to submit the task.

    Table 12 Task startup settings

    Parameter

    Description

    Start Time

    Set Start Time to Start upon task creation or Start at a specified time based on site requirements.

    NOTE:

    After a synchronization task is started, the performance of the source and destination databases may be affected. You are advised to start a synchronization task during off-peak hours.

  7. After the task is submitted, you can view and manage it on the Data Synchronization Management page.

    • You can view the task status. For more information about task status, see Task Statuses.
    • You can click in the upper-right corner to view the latest task status.
    • By default, DRS retains any task in the Configuration state for three days. After three days, DRS automatically deletes background resources, but the task status remains unchanged. When you configure the task again, DRS applies for resources for the task again. In this case, the IP address of the DRS instance changes.