Constraints

• Cold and hot data separation is unavailable for single-node instances, serverless instances, multi-primary instances, instances in a RegionlessDB cluster, or instances with TDE or cross-region backup enabled.
• After cold and hot data separation is enabled, cross-region backup and manual backup are not supported.
• Cold and hot data separation cannot be disabled after being enabled.
• Restoring data from an instance with cold and hot data separation enabled to a new instance involves copying the cold data from the bucket, leading to a longer restoration duration.

• Cold table operations are unavailable for frozen DB instances.
• Cold tables can only be queried using SELECT statements and the query speed is slow. Do not configure tables storing frequently updated data as cold tables.
• To prevent a creation failure, do not run DDL or DML statements on the selected table.
• Cold tables can only be queried and the query is slow. They do not support DDL or DML statements. You are advised to convert rarely accessed tables that store archived data to cold tables.
• If there is a foreign key in a cold table, DDL statements cannot be executed on the primary table corresponding to the foreign key.

• Cold data archiving can only be enabled for cluster instances.

• Temporary tables and views cannot be archived.
• The first partition of a partitioned table cannot be archived.
• HASH, KEY, LINEAR HASH, LINEAR KEY, or LIST DEFAULT HASH partitions cannot be archived.
• Subpartitions or partitions with subpartitions cannot be archived.
• System catalogs cannot be archived.
• Tables with full-text indexes cannot be archived.
• If the table to be archived is a partitioned table, only one partition can be archived at a time. If multiple partitions need to be archived, archive each partition separately.
• If data is archived by partition, DDL statements can only be executed on partitions other than the archived partition. The first partition does not support DROP. REMOVE PARTITIONING is not supported. Table-level DDL statements are not supported.
• If a SQL command is executed for archiving, special characters in the database name, table name, and partition name need to be escaped.
• The bucket for storing archived cold data uses a single-AZ parallel file system. If there is an AZ-level fault, you may not be able to access the cold data.

After a cold table is migrated back, the cold data will still be stored in the bucket. You must manually delete the cold data on the console after the automated backup retention period expires to avoid ongoing billing for the cold data.

• Instances with cold and hot data separation enabled only support full data restoration. Table-level data restoration is not supported. Data can only be restored to a new or the original instance using backups or PITR.
• Only the structure of a cold table can be backed up. Cold data cannot be backed up or restored.
• Data of instances with cold and hot data separation enabled cannot be restored to existing instances using backups. However, data of existing instances can be restored to instances with cold and hot data separation enabled using backups.

During migration using DRS, if you want to retain cold table data of a source database, migrate cold tables back to the source database and then migrate the database using DRS. Otherwise, the cold tables will be migrated to empty tables of the destination database.