Data Replication Service supports now DDM into DDM

Synchronization objects

• Full synchronization supports the synchronization of data, table structures, and indexes.
• Only MyISAM and InnoDB tables can be synchronized.
• The source database cannot contain tables whose sharding keys are of the timestamp type.

Restrictions on the source database

Enable skip-name-resolve for the source sharded database to reduce the possibility of connection timeout.

Use restrictions

General

Allgemein

• During the task startup or full synchronization, you are not advised to perform DDL operations, such as deletion, on the source database. Otherwise, the synchronization may fail.
• During synchronization, do not modify or delete the usernames, passwords, permissions, or ports of the source and destination databases.
• During synchronization, do not modify the table structure to be synchronized in the source database.
• During the synchronization, do not change the sharding key of a table on the source DDM instance, or change an un-sharded or broadcast table to a sharded table, or change a sharded table to an unsharded or broadcast table.

Incremental synchronization

• The source database cannot be restored.
• Some DDL statements are supported, including ADD COLUMN, CREATE TABLE, CREATE INDEX, DROP INDEX, RENAME INDEX and ADD INDEX.
• Perform DDL operations on the same table or column during off-peak hours at an interval of more than 1 minute.

Troubleshooting

• If any problem occurs during task creation, startup, full synchronization, incremental synchronization, or completion, rectify the fault by referring to Troubleshooting.

Other restrictions

• If the data types are incompatible, the synchronization may fail.
• If the source database contains non-standard floatingpoint data and the data can be written in loose mode but cannot be written in strict mode, there may be data inconsistency during synchronization.
• If a physically generated column in a table is generated based on a time type, the data in the column may be inconsistent.
• The destination table can contain more columns than the source table. However, the following failures must be avoided:

• Wenn die Datentypen nicht kompatibel sind, kann die Synchronisierung fehlschlagen.
• Wenn die Quelldatenbank nicht-standard-gleitende Dezimalwerte enthält und die Daten in lose Modus geschrieben werden können, aber nicht in strenger Modus, kann es bei der Synchronisierung zu Dateninkonsistenzen kommen.
• Wenn eine physisch generierte Spalte in einer Tabelle auf der Basis eines Zeit-Typs generiert wird, können die Daten in der Spalte inkonsistent sein.
• Die Ziel-Tabelle kann mehr Spalten enthalten als die Quelldatei. Es müssen jedoch die folgenden Fehler vermieden werden:

– Assume that extra columns on the destination cannot be null or have default values. If newly inserted data records are synchronized from the source to the destination, the extra columns will become null, which does not meet the requirements of the destination and will cause the task to fail.

– Assume that extra columns on the destination must be fixed at a default value and have a unique constraint. If newly inserted data records are synchronized from the source to the destination, the extra columns will contain multiple default values. That does not meet the unique constraint of the destination and will cause the task to fail.

• After a task is created, you cannot add schemas to the source database or modify the old schema to associate with the new RDS DB instance. Otherwise, data cannot be synchronized or the task fails.
• Create a schema for the destination database in advance.
• Ensure that the destination database is empty before starting the synchronization. Otherwise, data in the destination may be overwritten during incremental synchronization.
• If the destination instance uses columns of the TIMESTAMP or DATETIME data type as its sharding key, the seconds precision of the column is removed after the synchronization.
• The AUTO_INCREMENT value of a table in the destination cannot be less than that of a table in the source.
• During an incremental synchronization of table-level objects, renaming tables is not recommended.
• When editing a task, do not import a lot of data to the newly-added table. You are advised to edit the task during off-peak hours.