Connectors

Relational databases

The databases your business already runs on — read them as they are.

Relational databases are where most enterprise data still lives, and they are rarely the systems anyone wants to modernise first. BARRY connects to them as they are: no upgrade project, no agent on every database server, and no requirement that the version be recent. Extraction is read-only and deliberately low-impact, so production keeps doing its day job while data moves. Where the engine supports it you can go beyond scheduled full loads to watermark-based incremental loads or true change data capture, reading the database's own transaction log. Several of these connectors work in both directions, so the same system can be a source in one pipeline and a destination in another.

SQL Server / Azure SQL DBSourceSink

Reads SQL Server 2016 and later, Azure SQL Database, Managed Instance and RDS, with change tracking for change data capture. As a destination it bulk-loads straight into your tables.

PostgreSQLSourceSink

Reads any modern PostgreSQL, including managed services, and captures inserts, updates and deletes through logical replication. Writes use the binary COPY path, the fastest way into Postgres.

MySQLSourceSink

Reads MySQL and captures changes from the binary log using either GTIDs or file positions, so a table stays current without repeatedly scanning it.

MariaDBSourceSink

Wire-compatible with the MySQL connector, with support for MariaDB's three-part GTIDs so replication-based change capture works the same way.

Oracle DatabaseSource

Reads Oracle through the managed .NET driver using Easy Connect strings. Full and watermark-based incremental loads are supported; log-based change capture is not part of this connector.

IBM Db2Source

Reads Db2 through the platform-native driver, with separate Windows and Linux builds. Full and watermark-based incremental loads are supported.

ODBC (generic)Source

The fallback for anything with an ODBC driver on the client machine — SAP HANA, Sybase, Teradata, Informix and similar. Slower and with lower type fidelity than a native connector, but it reaches systems nothing else does.

ClickHouseSink

Writes into ClickHouse Cloud, open-source ClickHouse and hosted variants by streaming rows straight into an INSERT over the HTTP interface.

Apache DorisSink

Loads into Apache Doris through its Stream Load endpoint. Append is the supported write mode, which suits event-style and log-style tables.

Looking for something else?

Every connection is Source, Sink, Trigger or Alert — derived from the connector itself, never set by hand. Browse the full catalogue, or ask us about a system you don’t see.

Ready to unlock your data?

See how BARRY brings your on-premises data to the cloud — safely, and on your terms.