Sign InTry Free

Merge and Migrate Data from Sharded Tables

This document introduces the sharding support feature provided by Data Migration (DM). This feature allows you to merge and migrate the data of tables with the same or different table schemas in the upstream MySQL or MariaDB instances into one same table in the downstream TiDB. It supports not only migrating the upstream DML statements, but also coordinating to migrate the table schema change using DDL statements in multiple upstream sharded tables.

Overview

DM supports merging and migrating the data of multiple upstream sharded tables into one table in TiDB. During the migration, the DDL of each sharded table, and the DML before and after the DDL need to be coordinated. For the usage scenarios, DM supports two different modes: pessimistic mode and optimistic mode.

The pessimistic mode

When an upstream sharded table executes a DDL statement, the migration of this sharded table will be suspended. After all other sharded tables execute the same DDL, the DDL will be executed in the downstream and the data migration task will restart. The advantage of this mode is that it can ensure that the data migrated to the downstream will not go wrong. For details, refer to shard merge in pessimistic mode.

The optimistic mode

DM will automatically modify the DDL executed on a sharded table into a statement compatible with other sharded tables, and then migrate to the downstream. This will not block the DML migration of any sharded tables. The advantage of this mode is that it will not block data migration when processing DDL. However, improper use will cause migration interruption or even data inconsistency. For details, refer to shard merge in optimistic mode.

Contrast

Pessimistic modeOptimistic mode
Sharded tables that executes DDL suspend DML migrationSharded tables that executes DDL continue DML migration
The DDL execution order and statements of each sharded table must be the sameEach sharded table only needs to keep the table schema compatible with each other
The DDL is migrated to the downstream after the entire shard group is consistentThe DDL of each sharded table immediately affects the downstream
Wrong DDL operations can be intercepted after the detectionWrong DDL operations will be migrated to the downstream, which may cause inconsistency between the upstream and downstream data before the detection

Was this page helpful?

Download PDFRequest docs changesAsk questions on Discord
Playground
New
One-stop & interactive experience of TiDB's capabilities WITHOUT registration.
Products
TiDB
TiDB Dedicated
TiDB Serverless
Pricing
Get Demo
Get Started
© 2024 PingCAP. All Rights Reserved.
Privacy Policy.