Oracle Streams

pre-11.2 · replication · deprecated 12.1 · desupported 19c

Survived, partially.

✓ certified by the mortician

In life

Oracle Streams was the database's built-in framework for log-based capture, propagation and apply of changes. Capture processes mined redo, staged logical change records in Advanced Queuing queues, and apply processes replayed them on the same or another database. It was used for replication, data sharing between systems and change feeds, and was configured through DBMS_STREAMS_ADM and related packages.

Migrations care because a Streams configuration rarely shows up in application code; it appears as capture and apply processes, queue tables and rules. If downstream systems depend on a Streams feed, the Azure design needs a replacement pipeline, whether GoldenGate, Azure Data Factory or another change data capture tool, and the Azure database targets cannot participate in Streams at all.

Oracle's 12.1 Upgrade Guide deprecates Streams and names GoldenGate as the replacement. The 19c Upgrade Guide desupports it starting with 19.1 (notice dated April 2019), noting that Streams never supported 12c-era features such as multitenant. The same notice states that Advanced Queuing itself is not deprecated and remains fully supported.

Cause of departure

Desupported in Oracle Database 19c after deprecation in 12.1. Oracle names Oracle GoldenGate as the successor.

Obituary

Oracle Streams, born before 11.2, captured redo and continuously propagated database changes. It served as Oracle’s replication engine with quiet persistence.

Deprecated in 12.1 and desupported in 19c, Streams found Azure offered no direct reprieve. Azure Database Migration Service provides no online Oracle-source migration, while Azure Data Factory reduces deltas to customer-managed batch watermarks.

It is survived by Azure Database Migration Service, not at all; Azure Data Factory, only by workaround; Microsoft Fabric, partly; and Oracle Database@Azure, partly. GoldenGate is the Oracle replacement for active-active replication, though it is not included and may bring extra licensing costs.

Survived by

Microsoft Fabric partial effort M

Applies to Oracle 11.2, 12.1, 12.2, 18c

Fabric mirroring for Oracle reads the redo stream through LogMiner and lands changes as Delta tables in OneLake with latency as low as tens of seconds, which reproduces the capture-and-apply side of Streams for analytics consumers. It is one-directional into OneLake, requires archive log mode, supplemental logging and an on-premises data gateway, and cannot replicate into another Oracle or SQL database, so operational database-to-database replication still needs GoldenGate or application-level replication.

Enable archive log mode and supplemental logging on the source and grant the LogMiner-related privileges.
Install the on-premises data gateway with network reach to the Oracle listener.
Create a mirrored database in Fabric selecting the schemas or tables to replicate.
Point reporting and downstream pipelines at the SQL analytics endpoint or Delta tables.

Complications

major Fabric mirroring for Oracle requires LogMiner, archive log mode and supplemental logging on the source, an on-premises data gateway, and a database open in write mode; it supports Oracle 10 and later on-premises, on Azure VMs, on OCI, on Oracle Database@Azure and on Exadata.

· microsoft Mirror Oracle databases in Microsoft Fabric

Oracle Database@Azure partial effort M

Applies to Oracle 11.2, 12.1, 12.2, 18c

Streams was desupported in 19c, so on Oracle Database@Azure the Oracle-native successor is GoldenGate, which Microsoft lists as a monitored deployment type in the service and describes as the option for active-active replication. GoldenGate is not included in the Database@Azure price and needs its own licence, and its capture/replicat configuration differs from Streams rules.

Rewrite Streams capture, propagation and apply rules as GoldenGate extract, pump and replicat parameter files.
Deploy GoldenGate on Database@Azure or on Azure VMs with network reach to both endpoints.
Monitor GoldenGate deployments through Azure Monitor metrics and logs.

Complications

major Microsoft's Oracle Database@Azure guidance says GoldenGate is not included in the service and might incur extra licensing costs, while positioning it as the tool for active-active replication with application-level conflict handling.

· microsoft Business continuity and disaster recovery for Oracle Database@Azure (Exadata Database Service)

Azure Data Factory workaround effort L

Applies to Oracle 11.2, 12.1, 12.2, 18c

Oracle Streams captured redo and propagated changes between databases continuously. Azure Data Factory's Oracle connector is a batch copy source: Oracle is not among the connectors with native change data capture in mapping data flows, nor among the auto-incremental connectors, so deltas from Oracle can only be extracted with a customer-managed watermark pipeline (lookup the last high-water mark, copy rows above it, write the new mark). That gives scheduled micro-batches, not log-based streaming, and no capture of deletes without a soft-delete column.

Identify a monotonically increasing column (timestamp or sequence) on each replicated table, or add one.
Build a watermark pipeline: Lookup old mark, Copy activity with a WHERE on the column, stored procedure to store the new mark.
Handle deletes with soft-delete flags or periodic full reconciliation.
Schedule with tumbling-window triggers at the latency the consumers tolerate.

Complications

major Azure Data Factory's native change data capture in mapping data flows lists SAP CDC, Azure SQL Database, SQL Server, Azure SQL Managed Instance, Azure Cosmos DB and Snowflake as supported sources, and its auto-incremental extraction list also excludes Oracle, so Oracle deltas require a customer-managed watermark pipeline; pipelines are batch and only the CDC factory resource runs continuously.

· microsoft Change data capture in Azure Data Factory and Azure Synapse

Azure Database Migration Service none effort L

Applies to Oracle 11.2, 12.1, 12.2, 18c

Azure Database Migration Service has no online (continuous sync) scenario with Oracle as a source; the only Oracle scenarios are offline migrations to Azure SQL targets through SSMA. It therefore cannot stand in for Streams either as a replication engine or as a minimal-downtime migration path.

Plan Oracle migrations to Azure SQL as offline cutovers with SSMA or DMS-via-SSMA.
Use Fabric mirroring, Data Factory watermark pipelines or Oracle GoldenGate for continuous change movement.

Complications

blocker The Azure Database Migration Service scenario table marks Oracle as 'No' for online (continuous sync) migration to Azure SQL Database, Managed Instance and SQL VM, and lists no Oracle source at all for PostgreSQL targets.

· microsoft Azure Database Migration Service: supported migration scenarios

Notices of correction

oracle Deprecated and Desupported Features for Oracle Database 12c (12.1) · Upgrade Guide 12.1, E41397 · read 2026-09-26
oracle Oracle Database Changes, Desupports, and Deprecations (Upgrade Guide 19c) · Upgrade Guide 19c, E96252-82, Aug 2026 · read 2026-09-26
microsoft Mirror Oracle databases in Microsoft Fabric · page updated 2026-05-21 · read 2026-09-26
microsoft Mirroring in Microsoft Fabric · page updated 2026-08-28 · read 2026-09-26
microsoft Overview: Oracle Database@Azure · page updated 2026-09-15 · read 2026-09-26
microsoft Business continuity and disaster recovery for Oracle Database@Azure (Exadata Database Service) · page updated 2026-08-27; carries a deprecation notice (removal planned 2026-10-30) · read 2026-09-26
microsoft Change data capture in Azure Data Factory and Azure Synapse · page updated 2026-08-05 · read 2026-09-26
microsoft Copy data to and from Oracle with Azure Data Factory · page updated 2026-08-05 · read 2026-09-26
microsoft Azure Database Migration Service: supported migration scenarios · page updated 2026-02-20 · read 2026-09-26

← All notices