Capabilities of real-time database synchronization tasks

更新时间:
复制 MD 格式

DataWorks Data Integration provides a powerful real-time database synchronization solution that replicates all or selected tables from a source database to a destination data store with low latency in an automated, integrated full-and-incremental manner. Built on a real-time computing engine, this feature automatically performs the initial full data load and then seamlessly switches to continuous capture of incremental change data (CDC). It serves as a one-stop solution for use cases such as real-time cloud migration and real-time data warehouse ODS layer construction.

Use cases

  • Build a real-time ODS layer for your data warehouse

    Synchronize data from operational databases such as MySQL and Oracle to a real-time data warehouse like Hologres or StarRocks, providing fresh data for dashboards and ad hoc queries.

  • Real-time database replication for disaster recovery

    Establish a real-time replication link between two database instances for read/write splitting, read-only replicas, or disaster recovery across homogeneous or heterogeneous databases.

  • Real-time data migration to the cloud

    Seamlessly migrate databases from an on-premises data center to a cloud database service.

  • Build a real-time data lake or data middle platform

    Collect real-time change data from multiple operational databases into a central data lake (OSS or DLF) or data warehouse (MaxCompute or Hologres) to build a unified real-time data middle platform.

Core capabilities

Real-time database synchronization provides the following capabilities:

image

Core capability

Feature

Description

Synchronize entire databases between heterogeneous data sources

-

Real-time database synchronization supports migrating data from on-premises data centers or other cloud platforms to a data warehouse or data lake such as MaxCompute, Hologres, or Kafka. For more information, see Supported data sources and synchronization solutions.

Synchronize data in complex network environments

-

Real-time synchronization supports data from Alibaba Cloud database services, on-premises data centers, self-managed databases on ECS, and databases on other cloud platforms. Before configuration, ensure network connectivity between the resource group and the source and destination. For more information, see Network connectivity solutions.

Synchronization scenarios

Full synchronization

Performs a one-time full data synchronization from the source to the destination tables.

Incremental synchronization

Continuously captures streaming data from message queues or CDC logs and writes it to destination tables or specified partitions in real time.

Integrated full and incremental synchronization

  • Automatic full initialization: When a task starts for the first time, it automatically reads all existing data from the source database tables and writes it to the destination.

  • Seamless switchover to incremental sync: After the full synchronization phase is complete, the task automatically and seamlessly switches to CDC mode, continuously capturing inserts, deletes, and updates from the source and synchronizing them to the destination with millisecond-level latency.

Task configuration

Batch table synchronization

Synchronize all tables in a database, or use selection or filter rules to specify a subset of tables to synchronize.

Automatic table creation

Configure once to process hundreds of tables from the source database. The system automatically creates table schemas in the destination without manual intervention.

Flexible mapping

Supports custom naming rules for destination databases and tables, as well as custom field type mappings between source and destination, flexibly adapting to the data structure model of the destination.

DDL change detection (supported for certain synchronization links)

When the source table schema changes (such as creating or dropping tables or columns), you can configure the synchronization task to adopt one of the following response strategies:

  • Normal processing: The destination automatically applies the corresponding schema changes.

  • Alert: The changes are not applied. Only an alert notification is sent, and manual intervention is required.

  • Error: The task is immediately stopped and its status is set to Error.

DML rule configuration

DML message processing allows you to perform fine-grained filtering and control on the change data captured from the source (Insert, Update, Delete) before the data is written to the destination. You can use this rule to define the final processing strategy for different types of data change operations.

Dynamic partitioning

If the destination table is a partitioned table, dynamic partitioning is supported based on source fields or the event change time on the source side.

Important

Too many partitions degrade synchronization performance. If more than 1,000 partitions are added in a single day, partition creation fails and the task is terminated.

Task OM

Online intervention

Supports checkpoint-based resumption, which resumes execution from a specified time checkpoint after a task interruption to ensure no data loss during synchronization. Also supports re-running tasks for data backfill, anomaly repair, or logic change verification, ensuring data consistency and business continuity.

Monitoring and alerting

Supports monitoring rules for business latency, task status, failover, and DDL notifications, and supports alerting on triggered rules.

Resource tuning

DataWorks Data Integration is based on serverless resource groups and provides elastic scaling at the task level.

In addition, you can configure time-based elastic scaling policies to assign different resource specifications for the task at different times, such as during business peaks and off-peak hours.

DI Agent

AI Native

DI Agent is a vertical capability set of Data Agent for the data integration scenario. It covers six major capabilities: conversational task configuration, multimodal data processing, ChatDB data Q&A, intelligent diagnostics, periodic inspections, and IM channel integration.

Get started

To create a real-time database synchronization task, see Create a real-time database synchronization task.

To create and manage real-time database synchronization tasks through natural language conversations, you can use the Data Agent AI Native capability. Data Agent supports all task types, including single-table offline, single-table real-time, database-wide offline, and database-wide real-time tasks. You do not need to manually fill in configuration forms. A single sentence can complete the entire process from intent understanding to task deployment. For more information, see Use Data Agent to create data integration tasks.

Supported data sources

Source data sources

Destination data sources

MaxCompute

AnalyticDB for MySQL (V3.0)

ApsaraDB for OceanBase

Data Lake Formation (DLF)

DataHub

Doris

Elasticsearch

Hologres

Kafka

  • MySQL

  • MySQL (sharded)

  • PolarDB (sharded)

LogHub

OSS

OSS-HDFS

SelectDB

StarRocks

Lindorm