Data Transfer release notes

20262026

JuneJune

• Supported replication of data from Apache Kafka® and YDS queues to the Apache Iceberg™ data sink.

• The Object Storage data sink has been updated and is now available to all users.

MayMay

New features and improvementsNew features and improvements

• Added support for a new Apache Iceberg™ target endpoint.

• In Greenplum® endpoints, added support for Apache Cloudberry™.

AprilApril

New features and improvementsNew features and improvements

• Added support for transfer of non-table Greenplum® objects (functions, sequences, indexes).

• Added support for compression at the data format level for Parquet when writing to an Object Storage target.

• Added support for writing from Data Streams to OpenSearch.

MarchMarch

New features and improvementsNew features and improvements

• Supported the nginx data format for the Object Storage data source.

• Revised the name format for objects created in a Object Storage target: part--<hash(partID)>..<file_type>.

  • : timestamp by operation ID.
  • <hash(partID)>: UUID generated based on partID.
  • : Incrementing counter.
  • <file_type>: File extension (csv, parquet, json, proto, or nginx).

• For Snapshot transfers from PostgreSQL sources, supported permanent data reads from a parent table provided it is included in the endpoint or transfer table list.

• For YDB, Apache Kafka®, and YDS sources, supported a new data conversion type, Raw-to-table, which allows delivering raw data from a message queue.

FebruaryFebruary

New featuresNew features

• Significantly improved the speed of PostgreSQL-to-PostgreSQL data transfers.

• Improved database schema migration for PostgreSQL-to-PostgreSQL transfers. Entities are now migrated sequentially based on dependencies between them.

• Number of VMs allocated for the copy stage will be limited to 64.

Fixed issuesFixed issues

• Improved transfer stability for Apache Kafka® sources.
• Fixed sending metric to Monitoring from the Oracle source.
• Fixed the issue with PostgreSQL-to-PostgreSQL data transfers, where defining the table list at the transfer level (rather than the endpoint) caused the transfer of the entire schema.
• Fixed an issue where simultaneous updates to endpoint and transfer settings via the API and Terraform triggered multiple transfer restarts.
• Fixed the table vacuuming issue for the Managed Service for YDB target.
• Fixed freezes and duplicate delivery of large data volumes during replications from Managed Service for YDB.
• Fixed an issue where deletions failed to propagate to ClickHouse® if columns contained certain types, e.g., UUID.

JanuaryJanuary

You can now configure Apache Kafka® connections and Data Streams in the Schema Registry data source settings.