Data Transmission Service (DTS) is an Alibaba Cloud service for moving and processing data across databases — whether you're migrating to the cloud, keeping databases in sync across regions, or streaming real-time changes to downstream systems. It handles homogeneous and heterogeneous databases alike, across on-premises, Alibaba Cloud, and third-party cloud environments.
Core capabilities
DTS provides five functions that cover the full data movement lifecycle:
| Function | What it does | Typical use cases |
|---|---|---|
| One-time transfer of data from a source to a destination, with schema transformation support for heterogeneous databases. Keeps the source fully operational throughout, reducing application downtime to minutes. | Cloud migration, database relocation | |
| Continuous, real-time replication between two live databases — unidirectional or bidirectional. Includes a Data Delivery sub-feature for SDK-based delivery of incremental data (such as log data) to a target destination. | Active-active geo-redundancy, geo-disaster recovery, cross-border synchronization, real-time data warehousing, read/write splitting | |
Captures incremental changes (INSERT, UPDATE, DELETE) from a source database in real time and exposes them as a structured stream for downstream consumers. | Cache updates, asynchronous decoupling, ETL pipelines for heterogeneous sources | |
| Applies real-time ETL — filtering, field mapping, data masking, format conversion — to data as it flows through the data transmission process. | Data cleansing, sensitive data handling, schema normalization | |
| Compares source and destination data at the row level and generates a detailed report of any discrepancies. Runs either as a standalone task or alongside an active migration or synchronization task. | Post-migration verification, ongoing health monitoring of long-term sync tasks |
How DTS moves your data
For migration and synchronization tasks, DTS uses a two-phase approach — illustrated here for a typical migration:
Full load — DTS copies a complete snapshot of the source data to the destination.
Incremental sync — While the full load runs, DTS captures all ongoing changes using source database logs (such as binary log (Binlog) for MySQL or write-ahead log (WAL) for PostgreSQL) and applies them to the destination after the snapshot completes.
This two-phase approach keeps the source database fully operational throughout the process. Once the destination catches up with the source, you switch traffic — typically within minutes.
Data Synchronization tasks follow the same model but run the incremental phase continuously to maintain an always-current replica.
Key advantages
Broad compatibility
DTS connects relational databases (MySQL, PostgreSQL, SQL Server, Oracle), NoSQL stores (MongoDB, Redis), and big data warehouses (AnalyticDB, MaxCompute). For the full list, see Supported databases.
It also supports cross-environment topologies — on-premises to Alibaba Cloud (ApsaraDB RDS, PolarDB), cloud to cloud, and hybrid configurations — over public networks, Express Connect, VPN Gateway, or Smart Access Gateway (SAG). For cross-account scenarios, configure tasks using RAM authorization.
High performance
Full migration: up to 70 MB/s throughput with multi-threaded compression to reduce bandwidth usage
Real-time sync: up to 30,000 requests per second (RPS) with transaction-level concurrency
Reliable and secure
The DTS service cluster automatically fails over on node failures, ensuring continuous availability. DTS verifies data integrity around the clock. All transmissions use encrypted protocols, secure token authentication, and SSL encryption. For cross-border and cross-region synchronization, dedicated network connections provide additional isolation.
Simple to operate
Create and monitor tasks through a visual wizard in the DTS console. If a network or system exception interrupts a task, DTS resumes automatically from where it left off (breakpoint resumption) — no manual intervention needed.
Use cases
Zero-downtime migration to the cloud
Use Data Migration to move on-premises or self-managed ECS databases to ApsaraDB RDS or PolarDB. The two-phase approach (full load + incremental sync) keeps the source running continuously, reducing application switchover to minutes.
Active-active geo-redundancy and disaster recovery
Use Data Synchronization to build a geo-redundant architecture with bidirectional sync between databases in different regions. During an outage, traffic fails over rapidly to the standby region, maintaining business continuity.
Real-time data warehouse and cache updates
Use Data Subscription to stream database changes to AnalyticDB, ClickHouse, or Redis. DTS captures source logs and delivers them as a structured JSON stream, letting your application write directly to the data warehouse or update the cache without additional middleware.
In-flight data cleansing and formatting
Use Data Transformation to process data as it moves — for example, masking sensitive fields or merging source columns into a new destination field during synchronization. This eliminates the need for a separate transformation layer on the destination side.
Data consistency validation
Use Data Validation after a major migration or throughout a long-running synchronization task to confirm data accuracy. It runs full or range-based comparisons and identifies exactly which tables and rows have discrepancies.
Get started
Identify the DTS function that matches your scenario (see What's the difference between Data Migration and Data Synchronization? if you're unsure).
Confirm that your database is on the supported databases list.
Create a task in the DTS console or via the API or SDK:
What's next
Product architecture — system design and internal concepts
Specifications — performance specs for different task types
Basic concepts — terminology reference
DTS Insight — intelligent O&M assistant that monitors task status and assists with troubleshooting
FAQ
What's the difference between Data Migration and Data Synchronization?
| Data Migration | Data Synchronization | |
|---|---|---|
| Purpose | One-time move; source is typically retired after migration | Continuous replication; both sides stay active |
| Duration | Runs once, then stops automatically | Runs continuously |
| Advanced features | Schema transformation for heterogeneous sources | Bidirectional sync, conflict resolution |
| Choose when | Cloud migration, database replacement, data center relocation | Active-active geo-redundancy, read/write splitting, long-term cross-region replication |
In short: use Data Migration when the source is being retired; use Data Synchronization when both sides remain active.
How does DTS billing work?
DTS charges are based on your task type, covering task instance fees and public network/data traffic fees. For a detailed breakdown, see Billable items.