Microsoft Fabric Updates Blog

Accelerate AI on Oracle Databases with Open Mirroring, Fabric Data Agent, and Azure AI Foundry

Additional contributors: Venkat Ramakrishnan, Amir Jafari, Wilson Lee, Maraki Ketema

As organizations accelerate their hybrid cloud adoption strategies, Oracle Database@Azure has emerged as a critical platform for running Oracle database workloads using Exadata, Autonomous, Exadata Exascale and Base databases for the enterprise. However, deriving real-time, AI-powered insights in hybrid settings has long been challenging due to:

Latency and complexity: Connecting to Oracle Databases and data ingestion requires intensive setup, configuration, customization and ongoing maintenance.
Lack of native Oracle change data ingestion into Microsoft Fabric: No streamlined solution to bring Oracle change data directly into OneLake in Microsoft Fabric.
High operational overhead: Dependence on third-party tools and manual ETL processes drove up both cost and complexity.

Recent advancements in Microsoft Fabric and Azure AI Foundry are helping Oracle Database@Azure customers tackle these challenges. With Open Mirroring in Microsoft Fabric – a powerful feature that extends Mirroring in Fabric based on open Delta Lake table format – and the integration of Oracle (OCI) GoldenGate 23ai, customers can now replicate data from Oracle Databases into OneLake in Microsoft Fabric achieving low-latency, high-throughput replication without the need for complex ETL pipelines. Additionally, Oracle (OCI) GoldenGate 23ai helps unlock moving data from enterprise data sources (for example, DB2, Amazon Aurora) that support change data capture into OneLake in Microsoft Fabric.

Azure AI Foundry through its AI Agent Service now supports native integration with Microsoft Fabric data agent, enabling seamless access to Lakehouse, Warehouse, Power BI Semantic Models, and KQL databases. This empowers AI agents built using Azure AI Foundry to retrieve knowledge directly from Fabric for advanced data-driven insights and responses.

Together with Oracle GoldenGate’s integration into Open Mirroring in Microsoft Fabric and Azure AI Foundry’s AI capabilities, organizations can now achieve seamless, near real-time data replication, enable advanced AI capabilities drastically reducing engineering efforts and time to insight.

Unlocking AI-Powered capabilities in Microsoft Fabric

Once data is landed in OneLake, customers can build data agents, a native conversational AI capability in Fabric that allows users to interact with different data sources, including Oracle Database@Azure databases using natural language. This enables rapid development of AI-powered analytics, decision models, and business insights without writing complex SQL/DAX/KQL queries or requiring deep technical expertise. Fabric data agent has access to multiple organizational data sources — including Lakehouses, Warehouses, Power BI Semantic Models, and KQL databases. It interprets the natural language questions and user’s instructions, selects the most relevant data source(s), and generates structured queries to retrieve meaningful insights. As a result, data becomes more accessible across the organization, enabling faster, more informed decisions.

Integration overview: from Oracle to AI Insights

1. Data replication into OneLake in Microsoft Fabric via Open Mirroring

Oracle data is replicated in near real-time using Oracle GoldenGate integration into Open Mirroring in Microsoft Fabric. To setup the necessary replication from Oracle, please refer to the Oracle’s official documentation.

2. Shortcut to Lakehouse

A shortcut is established to link the mirrored data to a Lakehouse, serving as an effective way to enable a data agent in Microsoft Fabric.

3. Enable data agent in Microsoft Fabric

The data agent is then configured on the Lakehouse to process structured data and respond to natural language queries. Configuring the Fabric data agent involves specifying relevant tables from the lakehouse and providing AI instructions and example queries. Together, these elements enable the Fabric data agent to deliver more accurate and relevant results.

Below is a showcase of how a native Microsoft Fabric data agent is enabled on Oracle database@Azure ExaData sample Sales database mirrored into Microsoft Fabric. It highlights how Fabric data agent simplifies getting insights from your data in seconds, without the need to write complex SQL queries.

4. Integrate with Azure AI Foundry

For advanced AI workflows, the Fabric data agent can integrate with Azure AI Foundry as a knowledge source, enabling multi-agent communication. This integration securely grounds Azure AI agent outputs in enterprise knowledge, ensuring that responses are accurate, relevant, and contextually aware. In addition to structured data from Fabric, through Azure AI Foundry’s Agent service, the Azure AI agent can also leverage unstructured data sources, making it more intelligent and knowledge-driven.

Extending insights with Azure AI Foundry

Azure AI Foundry’s Agent Service enables the development of intelligent agents that aggregate data from multiple sources including Microsoft Fabric and external file systems. Leveraging its integration with Azure AI Search and Microsoft OneLake, users can include unstructured data such as documents, logs, and other knowledge bases in the agent’s reasoning.

The user can leverage two key methods for grounding AI agents with such unstructured data:

Direct OneLake integration: connect Fabric files via Azure AI Foundry’s Data + index features.
Azure AI search: index unstructured data through Azure AI Search and link it to the AI agent for enhanced retrieval and comprehension.

Once configured, the AI agent can seamlessly respond to queries by combining insights from structured and unstructured data sources.

Below is a showcase of an AI agent built in Azure AI Foundry using Fabric’s data agent and unstructured files (.csvs) indexed using Azure AI Search. This Agent is grounded with data mirrored from a sample Sales database running on Oracle Database@Azure Exadata workload into Microsoft Fabric and customer survey data on product feedback.

Licensing and pricing

Oracle Database@Azure: Oracle Database@Azure offers pricing at parity with Oracle Cloud Infrastructure (OCI), allowing customers to use existing Oracle licenses (BYOL), Unlimited License Agreements (ULAs), or purchase license-included options. Available through the Azure Marketplace, it supports Microsoft Azure Consumption Commitments (MACC) and qualifies for Oracle Support Rewards, helping reduce support costs. All Microsoft services remain available at standard pricing for all customers.
Microsoft Fabric: Mirroring in Microsoft Fabric offers free compute of the mirroring engine and free storage per tiered Fabric capacity. For more details, please refer to Mirroring cost details. For Fabric data agents, refer to data agent consumption for further details.
Azure AI Foundry: pricing depends on the compute and usage tier selected. Refer to Azure AI Foundry pricing.

Organizations running Oracle Databases on Premise, on Azure IaaS, Oracle Database@Azure workloads on Exadata, Autonomous, Exascale, Base databases, any data source supported by Oracle (OCI) GoldenGate can now unlock real-time, AI-powered analytics with the combined strength of Oracle (OCI) GoldenGate integration into Open Mirroring in Microsoft Fabric, Fabric data agent and Azure AI Foundry. This solution simplifies the complexity of data replication, reduces integration costs, and enables natural language interactions with enterprise data accelerating smarter decision-making powered by AI across the business.

Get started today

Now is the time to modernize your Oracle environments with Oracle Database@Azure, delivering greater agility, AI-driven insights, and enterprise-grade security.

Click here to get started today and connect with your local Microsoft sales team.

We encourage you to try out these new features and let us know what you think: