Skip to main content
Docs
Feature Requests Onboarding Trust Center Status Page

Community

Connector Improvement: Oracle: Add support for Oracle Autonomous DB (GoldenGate?)

Planned
Chris Redekop User

We would like Fivetran’s Managed Data Lake to support CDC streaming from Oracle Autonomous Databases.

The main goals are:

  • Enable lower-latency data replication
  • Ensure that tables in “history mode” capture every revision

Currently, our understanding is that the only supported option for ADB is Teleport Sync. While this works for smaller datasets or workflows where higher latency is acceptable, it becomes a bottleneck for large datasets or near-real-time analytics requirements.

The ideal implementation could leverage Oracle’s native CDC stack, for example:
Oracle ADB → GoldenGate → OCI Streams → Fivetran Managed Data Lake

This would allow Fivetran to ingest continuous change data while maintaining full historical fidelity in Iceberg tables. We understand there may be other architectures that achieve similar results, but the key requirements are: low latency, full revision capture, and seamless integration into Fivetran’s Managed Data Lake.

Please sign in to leave a comment.

Comments

2 comments

Date Votes
  • Official comment
    Vinayak Agrawal Fivetranner

    Hi Chris,



    This is a valuable feature request. Support for low-latency CDC streaming from Oracle Autonomous Databases into Fivetran Managed Data Lake has been added to our feature improvements backlog.



    I am interested to learn more about your use case. Specifically, can you share more details about the volume of data you’re working with, your latency requirements, and any existing constraints or workflows that this improvement would address? Understanding this will help us refine potential solutions.



    We will keep the community updated on this thread as there is any progress on this request.



    Thanks,

    Vin

  • Chris Redekop User

    Thanks for adding this to the backlog - happy to share more detail.

    We have several existing products that we are planning to migrate to Oracle Autonomous Database. Among these, the heavier workloads will generate upwards of ~50 GB of CDC data per day.

    This is our "wish list" of what we would like to use the fivetran managed datalake for:

    AI-powered, agent-based data interaction:
    We plan to use the Managed Data Lake as a source for interactive AI agents that help users explore, query, and gain insights from their data. This use case depends on fresh data so that AI responses reflect recent changes and current system state.
    While model training itself can tolerate higher latency, the serving/interaction layer benefits from low-latency CDC combined with full historical data, allowing agents to reason over both recent events and long-term trends.

    In-product dashboards and reporting:
    These features depend on low-latency CDC ingestion so that customer-facing metrics reflect changes quickly. Multi-hour sync windows significantly reduce their usefulness, whereas near-real-time ingestion (low-minute latency) enables responsive, trustworthy dashboards.

    Audit and compliance trails:
    For auditability (in-product as well as for compliance), we really need change history rather than just periodic snapshots. Tables would generally be sync'd in history mode, and we need every revision captured with high fidelity. Low latency is also somewhat important here so audit data is available in-product shortly after changes occur.