All Products
Search
Document Center

Data Transmission Service:Terms

Last Updated:Mar 28, 2026

This page defines the terms used in Data Transmission Service (DTS) documentation.

Instance types

Regular instance (single-zone instance)

A DTS instance deployed in single-zone mode, where all databases run on servers in the same zone. If a server fails, DTS switches the instance to another healthy server in the same zone. If the entire zone becomes unavailable, the DTS instance fails.

All DTS instances are currently regular instances in single-zone deployment mode.

Self-managed database (self-managed DB)

A database connected to DTS where Alibaba Cloud Instance is not selected as the Access Method. Self-managed databases include third-party cloud databases, on-premises databases, and databases running on an Elastic Compute Service (ECS) instance.

Task types

A DTS task consists of one or more sub-tasks, each handling a specific phase of data movement. Sub-tasks are grouped into three types:

Synchronization task type

  • Schema synchronization tasks

  • Full data synchronization tasks

  • Incremental data synchronization tasks

Migration task type

  • Schema migration tasks

  • Full data migration tasks

  • Incremental data migration tasks

Verification task type

  • Full data verification tasks

  • Incremental data verification tasks

  • Schema verification tasks

Schema task (structure task)

The sub-task that runs Schema Synchronization or Schema Migration.

Full data task (full task)

The sub-task that runs Full Data Synchronization or Full Data Migration.

Incremental task

The sub-task that runs Incremental Data Synchronization or Incremental Data Migration.

By default, a synchronization instance includes Incremental Data Synchronization.

Data verification task (verification task)

A task that checks data consistency between the source and destination databases. Includes full data verification tasks, incremental data verification tasks, and schema verification tasks.

Synchronization and migration phases

DTS moves data through up to three sequential phases: schema, full data, and incremental data.

Schema synchronization (schema migration / structure migration)

The phase where DTS reads schema objects — including tables, views, triggers, and stored procedures — from the source database and writes them to the destination database.

For heterogeneous database pairs, DTS converts schema types automatically before writing. For example, DTS converts the NUMBER type in an Oracle database to the DECIMAL type in MySQL.

Full data synchronization (full synchronization / full migration)

The phase where DTS reads all historical data from the source database — data that existed before the DTS instance started — and writes it to the destination database.

Incremental data synchronization (incremental synchronization / incremental migration)

The phase where DTS captures and applies ongoing data changes while the task is running. DTS retrieves incremental change statements from the source database (such as statements from the MySQL Binlog), converts them to match the destination database type, and executes them on the destination. This enables real-time synchronization or migration without service disruption.

Important

Note the following behavior:

  • If the destination database is Kafka or RocketMQ, DTS writes data directly after each incremental change.

  • Incremental data synchronization and migration run continuously and do not stop automatically. To stop the process, manually stop or release the instance. For more information, see Terminate a DTS instance and Release DTS instances.

Initial synchronization

The setup phase that runs before DTS begins synchronizing incremental data. It consists of three steps, in order:

  1. Incremental data collection: Starts capturing all changes to the objects to synchronize from the source database.

  2. Initial schema synchronization: Copies the schema of each object from the source to the destination database.

  3. Initial full data synchronization: Copies the historical data of each object from the source to the destination database.

Precheck

The validation step that runs before a DTS instance starts. DTS checks:

  • Connectivity between the source and destination databases

  • Database account permissions

  • Binary log settings

  • Database versions

For more information, see Source database connectivity.

If a precheck fails, view the failure details, fix the reported issues, and run the precheck again.

Data operations

Data update

An operation that modifies data without changing the schema, such as INSERT, DELETE, or UPDATE.

Schema update

An operation that modifies the schema structure, such as CREATE TABLE, ALTER TABLE, or DROP VIEW.

Performance and latency metrics

Synchronization performance (migration performance)

The number of incremental data records synchronized or migrated to the destination database per second. Unit: records per second (RPS).

For instance class specifications, see:

Synchronization latency (migration latency)

The difference between the timestamp of the latest record written to the destination database and the current timestamp of the source database. A latency of zero means the destination is fully in sync with the source.

When no data is being written to the source database, the displayed latency may fluctuate within 2 seconds even if the actual value is 0.

Change tracking terms

Timestamp range

The time window of incremental data stored in a change tracking task. By default, DTS retains incremental data generated in the last 24 hours. DTS periodically removes expired data and updates the timestamp range accordingly.

The timestamp of each incremental data record reflects when the data was updated in the source database and written to the transaction log.

Consumption checkpoint (consumer offset)

The timestamp of the latest incremental data record consumed by a client. Each time the client processes a record, it sends a confirmation to DTS, which updates and saves the checkpoint. If the client stops responding, DTS automatically resumes from the last saved checkpoint and pushes data to another consumer.