In OceanBase AP scenarios, data transmission encompasses data migration, data import, data synchronization, and data subscription: each targeting database switchover and cutover, batch table loading, continuous write alignment, and change event distribution, respectively. The following section summarizes the key selection points based on business objectives and provides corresponding documentation links.
Data migration
Complete target database setup or replacement within a specified time point or phase, switch business read/write to OceanBase (or a new tenant), and achieve schema alignment and data convergence.
Typical scenarios: One-time or phased replacement from MySQL/Oracle, etc.; OceanBase version upgrade or cluster replacement; migrating TP databases first and then handling AP; requiring only historical snapshots at a certain moment without continuous CDC for a few offline relocations.
Related documents: Data migration overview, Use tools to migrate data, Data import strategies and best practices, Best practices for migrating mainstream AP databases to OceanBase.
Data import
When importing external data into an internal table of OceanBase, the data source can be local files, object storage, HDFS, etc. Import can be completed using methods such as LOAD DATA, external tables, the FILES table function, or INSERT INTO ... SELECT.
Typical scenarios: Batch warehousing of historical details and log archives; CSV/ORC/Parquet files located in remote paths such as OSS or HDFS; writing ultra-large tables or full data more efficiently using direct load; reading lakehouse data through external tables or
FILESand then importing it into internal tables.Related documents: Data import overview, LOAD DATA FROM FILES, LOAD DATA INFILE, LOAD DATA LOCAL INFILE, Direct load overview, Full direct load, INSERT INTO SELECT FROM FILES, Import using INSERT INTO SELECT from external tables, Import data using obloader.
Data synchronization
Maintain business-acceptable consistency or an upper bound on latency between the target database (often OceanBase) and the source database over a long period, with tools continuously writing to the target database.
Typical scenarios: Intra-city/out-of-city disaster recovery, dual-machine hot standby; short-period dual-write and rollback windows before and after a cutover; aligning replicas with the primary database under multi-active or read/write splitting.
Related documents: Data synchronization overview.
Data subscription
Continuously read database logs or CDC events from upstream, parse them, and transmit them to Kafka, Flink, etc., for consumption, processing, or selective database writing by multiple downstream systems.
Typical scenarios: Use components like Canal or Flink CDC to subscribe to database changes, and write the changed data to messaging systems such as Kafka for real-time data warehouses, stream processing tasks, or multiple business systems to consume separately.
Related documents: Data subscription overview, Data synchronization overview (read alongside the OMS database writing path), Data import strategies and best practices.