Skip to main content
cancel
Turn on suggestions
Auto-suggest helps you quickly narrow down your search results by suggesting possible matches as you type.
Showing results for 
Search instead for 
Did you mean: 

Data Days is here! Join us now for 60+ days of learning, challenges, and connection. Learn more

murray-kp
Microsoft Employee
‎06-02-2026 01:24 PM

Fabric June 2026 Feature Summary

Welcome to the June 2026 Fabric update!

 

This month’s release brings a wide mix of improvements across the platform—from core data engineering and warehouse capabilities to continued investments in data agents, real-time intelligence, and enterprise-ready governance. You’ll see a strong focus on making complex scenarios more manageable, whether that’s simplifying storage and lifecycle management in OneLake, improving observability and reliability for data agents, or enabling more scalable, secure integrations across your data estate.

As always, we’ve highlighted the updates that will have the biggest day-to-day impact, along with previews of what’s coming next so you can plan ahead.

 

 

 

 

 

Events and Announcements

A new Data Days event starts this month!

In 2026, we’re going bigger than ever. Data Days will now expand beyond Fabric and Power BI and include SQL and AI. We are planning over 100 live sessions, more than 5 contests and challenges, and dozens of study groups and learning opportunities. You aren’t going to want to miss this event.

It all starts on June 15, 2026. Pre-register to get updates on Data Days activities, voucher offers and more.

 

Europe’s largest Microsoft Fabric and SQL event lands in Barcelona, September 28 – October 1, 2026

Explore what’s possible with Microsoft Fabric and get up to speed on the latest in SQL, analytics, and AI.

 

From 130 sessions and 4 keynotes to workshops, the expo, community spaces, and the Power BI DataViz World Championships, this is where the data community comes together. Learn directly from Microsoft and community experts shaping the future of Fabric and SQL.

Register now and save €200 with code FABCMTY200.

 

Fabric Platform

Fabric Git Integration – GitHub Enterprise Cloud with Data Residency (Generally Available)

Recently, we introduced support for GitHub Enterprise Cloud with data residency (ghe.com) instances in Microsoft Fabric Git integration in preview. Now this capability is available, enabling organizations to store GitHub Enterprise data within specific geographic boundaries and meet regulatory and compliance requirements.

 

Figure: GitHub Enterprise cloud for data residency – settings.Figure: GitHub Enterprise cloud for data residency – settings.

 

 To learn more, refer to the Microsoft Fabric and GitHub Enterprise Cloud with Data Residency support documentation.

 

OneLake

Optimize your storage costs with OneLake storage tiers and lifecycle management (Preview)

While OneLake makes it easy to centralize your data estate, keeping long-term data can drive up costs over time. Admins need ways to control their storage costs without giving up historical data often required for compliance and auditing. With OneLake storage tiers and lifecycle management, it's never been easier to optimize your storage bill on OneLake by moving your historical data into cost-effective tiers automatically.

 

OneLake lifecycle management policies simplify tiering by automatically moving files between tiers with customizable rules. Rules can change tiers based on when a file was created, last modified, or last accessed. Once defined, policies run automatically in the background, making tier management simple and automatic.

 

Figure: A lifecycle rule can contain multiple actions and conditions and be scoped to paths within a workspace.Figure: A lifecycle rule can contain multiple actions and conditions and be scoped to paths within a workspace.

To learn more, refer to the OneLake storage tiers documentation.

 

Understand your storage with OneLake item-size reporting (Preview)

As data estates continue to grow, data admins and owners need the tools to make informed decisions about the cost and governance of their data. To address these challenges, OneLake item-size reporting gives workspace administrators visibility into storage usage across OneLake, including system and soft-deleted data. 

 

Single page, capabilities for admins:

  • See all items in a workspace storing data in OneLake. 
  • Sort and search for items by amount of data stored. 
  • Understand how storage breaks down across visible, system, and soft-deleted data. 
  • Refresh on demand with a single selection. 

Figure: The storage report summarizes the amount of data stored in each item in your workspace.Figure: The storage report summarizes the amount of data stored in each item in your workspace.

 

Previously, it was difficult to get an item-level storage breakdown. The Capacity Metrics app aggregates storage at the workspace-level, leading admins to manually investigate with tools like Azure Storage Explorer, which can’t view system data. OneLake Item-size Reporting closes that gap by reporting the amount of data stored in every item, including system and soft-deleted data, with a single selection.

To learn more, refer to the getting the size of OneLake items documentation.

 

Data Engineering

Extended Lineage for Materialized Lake Views

Materialized lake views already let you write SQL and PySpark transformations and let Fabric handle the refresh. Extended lineage takes that further — your materialized lake views now span lakehouse boundaries. You can define all your views in one lakehouse while referencing source tables and upstream views in other lakehouses, even across workspaces. Fabric resolves the dependencies and refreshes everything in the right order.

 

 

See everything from one place
Turn on the Extended lineage toggle and the lineage expands beyond the current lakehouse, showing every upstream view, source table, and shortcut all the way to the original data. Each node shows its name, lakehouse name, and type. Faulted nodes flag dependencies that are missing or inaccessible before you run anything.

 

 

Refresh everything from one place
Create a schedule that includes upstream lakehouses. Pick which lakehouses to include from a tree, set a cadence, and Fabric refreshes upstream views first, then downstream, in dependency order. Advanced settings let you pick the Environment and toggle Optimal refresh per schedule.

 

Three ways to run on demand:

  • Refresh without dependent lineage — refreshes only the materialized lake views you select.
  • Refresh with dependent lineage — includes upstream dependencies within the current lakehouse.
  • Refresh with extended lineage — includes upstream lakehouses, refreshing the full lineage in dependency order

To learn more, refer to the Schedule a materialized lake view refresh documentation.

 

 

Define Authorizer User data functions for API for GraphQL (Preview) 

Authorizer user data functions enable custom authorization logic for GraphQL APIs by evaluating authenticated request details before processing. This capability lets API owners enforce business-specific access rules and gives you more control. An Authorizer UDF evaluates details from the authenticated request and determines whether to allow the call. The authorization logic runs as a function that’s automatically invoked for each incoming API request.

 

Use an Authorizer user data function to:

  • Apply custom business logic before API execution.
  • Restrict access based on user identity or token claims.
  • Validate service principle separately from user accounts.

Figure: Enable custom authorization with a user data function for your GraphQL API.Figure: Enable custom authorization with a user data function for your GraphQL API.

 

Data Science

AI functions: New models, no package dependency, better usage stats (Generally Available)

Fabric AI Functions now use gpt-5-mini as the default model, with “low” reasoning enabled. This powers AI Functions across pandas, PySpark, Data Warehouse, and Dataflows Gen2. For more sophisticated transformations, users may configure gpt-5.1 or tune the reasoning_effort parameter for additional compute and higher-quality results.

 

We are retiring the older gpt-4.1 model series. Pipelines pinned to gpt-4.1 (retiring May 30) should migrate to gpt-5.1, and those pinned to gpt-4.1-mini (retiring June 15) should migrate to gpt-5-mini.

 

We’ve also simplified PySpark AI Function chaining. The PySpark .ai interface now stays bound to the result schema, so chains like summarize → classify no longer require intermediate DataFrames. In addition, PySpark now supports df.ai.stats for detailed token usage after any AI function call, including reasoning token breakdowns.

 

For pandas, AI Functions no longer require the openai-python package. Capacity-limited rows are surfaced as CapacityExceededResult, enabling clean retries via aifunc.split_results.

 

To learn more, refer to the AI Functions documentation. 

 

Conversational Analytics

Fabric data agents to support Service Principals (Preview)

Fabric Data Agents now support service principals, enabling developers to authenticate and run data agents through app identities instead of requiring delegated user credentials. This update unlocks more secure and scalable enterprise integration scenarios, including backend services, automated workflows, and custom applications built with tools like Microsoft Foundry, Copilot Studio, and MCP-based agent frameworks.

 

With service principal authentication, organizations can simplify governance, reduce dependency on interactive sign-ins, and better operationalize conversational analytics and AI-powered experiences on top of governed Fabric data sources.

To learn more, refer to the Service Principal Support for Data Agents in Fabric (Preview) blog post.

 

Observability for Fabric Data Agent in Microsoft Foundry (Preview)

If you are building agents in Microsoft Foundry, you already know that Fabric data agent can be connected as a tool to your Foundry agent. Getting an agent to work is only the first step. The challenge lies in keeping it working reliably as your data changes, as your users ask new types of questions, and as your environment evolves.

 

Foundry Observability already gives you a tracing and monitoring layer for your agents. Think of it as your window into what's happening when your agent runs. You can trace each request, see how long each step takes, and understand where things slow down or break.

Now, Fabric data agents show up in Foundry Observability. This means that once you add Fabric data agent as a tool to your Foundry agent, you get detailed telemetry for every call to it. You can see latency (how long the data agent took to respond), status (did it succeed or fail), and error details when something goes wrong. Agent builders in Foundry can now actually debug their Fabric data agent calls properly. Instead of guessing why a response was slow or wrong, you have concrete data.

 

For anyone running agents in production, this is essential. You need to know when things degrade before your users tell you. Foundry Observability now gives you that visibility for Fabric data agent.

 

Data agents in Microsoft 365 Copilot (Generally Available)

You can publish your Fabric data agents as a declarative agent in M365 Copilot, which means your users can ask questions about their data right where they already work.

 

We have further enhanced our experience with additional support for longer running operations. If a user asks a complex question that takes more time to process, the system handles it gracefully. No timeouts and no dropped requests. Complex queries are now fully supported and will return an answer even if they take a bit longer to compute.

 

Creator Agent Release for SQL and Eventhouse Sources in Data agent (Preview)

This release introduces an AI-assisted experience that helps users set up data agents by generating and refining configurations through an interactive, guided workflow, replacing manual setup. It addresses key customer pain points around confusion between Agent Instructions, Data Source Instructions, and example queries, as well as the difficulty of iterating configurations effectively.

 

 

This AI assisted creation mode acts as an intelligent assistant that collaborates with users to produce high-quality configurations based on schema, context, and conversation. It enables automatic generation of core components such as agent instructions, data source guidance, and for example queries, reducing the need for deep technical expertise. The experience is designed to be iterative, allowing users to validate, adjust, and improve configurations over time as new requirements or feedback emerge. It can also tackle ambiguity and inconsistencies across configurations by identifying and recommending improvements.

 

 

At this stage, the preview focuses on SQL and Eventhouse scenarios with a setup and test workflow where users review and accept AI-generated updates. You can expect more data sources to be supported over time! Overall, this feature significantly streamlines Data Agent onboarding and optimization by making configuration more intuitive, scalable, and reliable.

 

Preview Runtime Opt-In

The Preview Fabric Data Agent Runtime is a new feature that provides early access to improvements shaping the default behavior of the Data Agent. By enabling the preview runtime, users can experience updates to core tools and system behaviors such as enhancements to NL2SQL or higher query result limits—before they are broadly rolled out.

 

The preview runtime focuses on foundational, end-to-end changes in how the agent operates, rather than introducing optional features or modes. These updates are designed to mature into the default experience over time, giving users an opportunity to validate changes and provide feedback early, while model updates continue to be applied independently across both preview and standard runtimes as needed.

 

Figure: Users can switch between the Standard and Preview runtime.Figure: Users can switch between the Standard and Preview runtime.

 

To learn more, refer to the documentation.

 

Improved NL2SQL Engine (Preview)

Data agents now leverage a new natural language to SQL (NL2SQL) set of tools when you enable the Preview Fabric Data Agent Runtime (a new setting). This enhanced SQL generation mode brings significant improvements in accuracy, transparency, and resilience to real-world complexity. NL2SQL is the natural language to use query capability across Lakehouses, Warehouses, and Mirrored Databases, translating user questions into SQL queries; with this update, it better leverages example queries to follow query patterns, asks clarifying questions when intent is ambiguous, and can intelligently explore data when scenarios fall outside predefined examples.

 

It also improves reliability through smarter filter mapping and provides enhanced visibility into run steps, applied examples, and structured diagnostics, making it easier to understand, debug, and trust query results.

Figure: The new NL2SQL engine can examine the data more closely when generating a query.Figure: The new NL2SQL engine can examine the data more closely when generating a query.

To learn more, refer to the documentation.

 

Improved Data Agent Routing Across Sources

The Fabric Data Agent orchestrator has improved its ability to route queries to the right data source, especially in scenarios where multiple sources are configured. When user intent is ambiguous or spans multiple datasets, the agent can now more effectively reason over a larger set of signals—including schema, for example queries, and data source descriptions—to determine the most relevant source.

 

These improvements enhance routing accuracy across both SQL and KQL sources, helping ensure queries are directed to the right place and results are more reliable. To learn more, see the Fabric Data Agent configuration documentation.

Figure: The run steps highlight when the agent needs to inspect additional metadata for data source routing.Figure: The run steps highlight when the agent needs to inspect additional metadata for data source routing.

To learn more, refer to the documentation. 

 

Code Interpreter Tool (Preview)

Code Interpreter is now available as a tool within the Fabric Data Agent, enabling advanced analysis, transformation, and visualization directly within agent workflows. By executing Python, the agent can go beyond database queries to perform tasks like statistical analysis, forecasting, and generating rich Python visuals—making it easier to answer complex questions and produce more actionable insights.

 

For example: “Forecast next quarter’s revenue with confidence intervals,” “Create a cohort retention heatmap by signup month,” “Test if conversion rates differ significantly between campaigns,” or “Combine support tickets with product usage data to identify churn patterns.”

Figure: Users can view the code generated by code interpreterFigure: Users can view the code generated by code interpreter

 

To learn more, refer to the Code Interpreter documentation. 

 

Data Warehouse

GPU-Accelerated Fabric Data Warehouse

AI workloads and agentic experiences are fundamentally changing what users expect from their data platform. Teams are shifting from static dashboards to dynamic, application-driven interactions. Every query now sits in the critical path of a real-time experience. GPU-Accelerated Fabric Data Warehouse makes this shift possible.

 

 

By bringing GPU-powered execution directly into the data warehouse, we enable a new class of workloads. Agents, applications, and reports can continuously issue complex analytical queries. These experiences remain responsive, interactive, and seamless to the end user. The capability is simple to enable and built for enterprise scale and reliability. It removes the traditional tradeoffs between performance, concurrency, and complexity.

 

 

The result is not just faster analytics. It is a new foundation for AI-driven innovation. Teams can build smarter applications that reason over massive datasets in real time. They can deliver more interactive agent experiences. They can unify operational and analytical workloads on a single platform without friction.

short_gif_GPU.gif

 

CI/CD Support for SQL analytics endpoint (Preview)

SQL Analytics Endpoint now supports CI/CD using DacFx, enabling teams to manage schema changes with the same rigor and repeatability used in enterprise database DevOps.  Developers can control SQL Analytics Endpoint definitions as DacFx database projects alongside other Fabric items in Git and deploy incremental schema changes through Fabric Deployment Pipelines.

 

This brings predictable, safe, and automated schema deployments to the SQL Analytics Endpoint—aligning analytics development with modern DevOps practices and eliminating manual, error‑prone changes across environments.

Figure: Commit SQL Analytics Endpoint changes to source control.Figure: Commit SQL Analytics Endpoint changes to source control.

 

Pre & Postscripts for Fabric Warehouse Deployments (Preview)

Pre & Postscripts enable teams to automatically run controlled SQL actions before and after Fabric Warehouse deployments, reducing manual intervention and lowering deployment risk in CI/CD workflows.

  • Before deployment — Run SQL scripts to validate and prepare the target environment—such as checking prerequisites, verifying schema readiness, ensuring required objects exist, and confirming deployment conditions before schema changes are applied.
  • After deployment — Run SQL scripts to finalize the environment by applying permissions, seeding or transforming data, configuring settings, and performing post‑deployment validation or cleanup.
  • Outcome — Teams achieve repeatable, predictable, and low‑risk Fabric Warehouse deployments, with deployment logic fully integrated into CI/CD pipelines rather than relying on manual, error‑prone steps.

 

Figure: Setting a shared query as pre/post in a Fabric Warehouse.Figure: Setting a shared query as pre/post in a Fabric Warehouse.

 

 

ALTER COLUMN (Preview)

ALTER COLUMN in Microsoft Fabric Data Warehouse enables column definition changes such as changing column length, precision, and scale directly on existing tables using standard T-SQL syntax. Changes take effect instantly with no underlying data rewrite, keeping pipelines and reports running without disruption and reducing friction in CI/CD and migration workflows.

 

For these supported scenarios, teams can reduce reliance on CTAS based workarounds, simplifying operations, lowering overhead, and reducing deployment risk. Please refer to the documentation, ALTER TABLE (Transact-SQL), for the most up-to-date list of supported features and behavioral considerations.

 

Datawarehouse Monitor (Preview)

Datawarehouse Monitor (previously Query Activity) provides a single place to monitor running and historical queries, quickly identify expensive or failed queries using rich per-query metrics (such as CPU time, data scanned, and cache usage), and drill into query run history, all without writing T-SQL. This enables faster performance troubleshooting and immediate action, including the ability to cancel runaway queries.

 

image015.gif

Figure: Data Warehouse Monitor Experience.

 

To learn more, refer to the Monitor T-SQL queries (Preview) documentation to get started. 

 

New enhanced metadata sync in SQL analytics endpoint (Preview)

The new metadata sync for SQL analytics endpoints addresses data staleness, one of the most common customer complaints. We are delivering a 30-second data freshness SLO. Once delta logs reflecting a data change are available in storage, the SQL endpoint will return the latest data within 30 seconds—even if the endpoint was previously deactivated. This can be enabled for new SQL endpoints in new or existing workspaces using the “new enhanced metadata sync” workspace setting.

Figure: Workspace setting to enable the new metadata sync.Figure: Workspace setting to enable the new metadata sync.

 

To learn more, refer to the New metadata sync and more in SQL Analytics Endpoint (Preview) blog post or check out the SQL Analytics Endpoint Metadata Sync documentation.

 

Configurable Retention in Fabric Warehouse (Preview)

Fabric Warehouse now gives you full control over your data history retention window. Built on the power of Delta — where every insert, update, and delete is versioned and preserved — you can now configure your data retention period anywhere from 1 to 120 days, tailored to your workload, compliance, and storage needs. No more one-size-fits-all 30-day ceiling.

Figure: Configure data retention through T-SQL.Figure: Configure data retention through T-SQL.

If a warehouse is accidentally dropped, Dropped Retention ensures it's recoverable for up to 90 days if configured — keeping your data safe.

 

Whether you need a shorter window to optimize storage costs or a longer trial for audit and compliance, Fabric Warehouse puts the control in your hands. Try Configurable Retention today and tell us what you think. We built this because you asked — and we'd love to hear how it's working for you.

 

Time Travel using SQL analytics endpoint (Preview)

The SQL analytics endpoint is where you can query Lakehouse using simple T-SQL — no data movement, no copies. We're now extending the same Time Travel experience available in Fabric Data Warehouse to the SQL analytics endpoint, so you can query your Lakehouse data exactly as it looked at any prior point in time. Simply add the OPTION (FOR TIMESTAMP AS OF 'yyyy-MM-ddTHH:mm:ss[.fff]') clause once at the T-SQL statement level, and the point in time is consistently applied across the entire query — whether it's a simple SELECT, a complex multi-table join, a view, or a stored procedure. One clause, one timestamp, end-to-end consistent results — perfect for troubleshooting, audits, ETL validation, and reproducing historical reports, all within the Lakehouse you already work with.

Figure: Time travel at SQL Endpoint.Figure: Time travel at SQL Endpoint.

 

Real-Time Intelligence

Stream Mirrored Database change feeds into Eventstreams (Preview)

Customers using Extended Capabilities in Mirroring to capture Delta Change Data Feed (CDF) can now stream those row-level changes directly into Fabric Eventstreams for low-latency, event-driven processing.

 

The new Mirrored Database Change Feed connector provides a fully managed path from mirrored change feeds to real-time intelligence. This eliminates the need to write custom Spark notebooks to poll for incremental updates and use Eventstreams SQL and no-code operators to process changes and build event-driven applications using Activator.

  • Discover CDF-enabled mirrored databases in the Real-Time Hub and connect to an Eventstream in a few clicks — no code required.
  • Stream row-level inserts, updates, and deletes with full fidelity, preserving the source table structure and changing metadata.
  • Apply Eventstream processing — SQL operators, filtering, and aggregation — and route outputs to Eventhouse, Activator, Lakehouse, and other destinations simultaneously.

  • Works with all sources supported by Fabric Mirrored Databases, including Azure SQL, Cosmos DB, Oracle, PostgreSQL, Snowflake, and Open Mirroring partners.

 

Figure: Streaming mirrored database change events through Eventstreams for real-time processing and analytics.Figure: Streaming mirrored database change events through Eventstreams for real-time processing and analytics.

 

  • Prerequisite: Delta Change Data Feed must be enabled on your mirrored database through Extended Capabilities.

For more information about this feature, refer to the Building real-time, event-driven applications on Mirrored Database Change Feeds with Fabric Eventstr... documentation.

 

 

Real-Time Dashboards – Powered by AI, a new way to create visuals (Preview)

We've redesigned the tile editing experience for Real-Time Dashboards from the ground up with an AI-first approach that makes building live visualizations faster and more accessible.

 

The new editor removes barriers for business users and operations teams who need real-time dashboarding capabilities but may not feel comfortable in developer-first environments. Now, anyone can start by choosing a visual type and describing what they need in natural language—Copilot generates the visualization for review without requiring KQL expertise.

 

Figure: Use Copilot inline prompt to create a new visual in Real-Time Dashboard.Figure: Use Copilot inline prompt to create a new visual in Real-Time Dashboard.

 

The redesigned experience includes a larger preview area that reduces on-screen distractions, and fully adaptive panels that let you focus on your preferred workflow—whether that's chatting with Copilot, writing queries, reviewing results, or browsing the data schema. Every Copilot-generated iteration is preserved in a chat history maintained by visual and across sessions, giving you a safety net to revert to any previous suggestion as you refine your work. You can also switch seamlessly between AI-assisted and manual editing: apply a Copilot suggestion, then fine-tune the KQL or adjust formatting, colors, and labels as needed.

 

Figure: Use Copilot to refine and apply a visual in Real-Time Dashboard.Figure: Use Copilot to refine and apply a visual in Real-Time Dashboard.

 

The result is that teams can onboard more people into building dashboards, faster—while developers benefit from a cleaner, better-organized layout. The new experience is available now when editing existing Real-Time Dashboards or creating new ones.

To learn more, refer to the Real-Time Dashboard documentation.

 

Time Series Visualization (Preview)

Real-Time Dashboards now include dedicated capabilities for time series analysis, making it easier to navigate, compare, and customize visualizations of time-based data.

 

Time series datasets often contain dozens or hundreds of data series. This feature provides purpose-built tools to help you find patterns, compare values across time periods, and identify anomalies—without getting lost in the data.

 

Navigation

When working with multivariate time series data, finding the right series matters. Time Series Visualization includes a legend search bar that lets you quickly locate specific data series by name. You can focus on a single series by selecting it from the chart or legend, and the corresponding elements highlight automatically.

 

For datasets with logical groupings—like sensors organized by machine or metrics organized by region—you can navigate through groups using simple Next and Prev controls. This makes it easy to step through your data systematically without manually clicking through dozens of legend items.

 

Comparison

Identifying trends and anomalies often requires comparison. You can now pin a data series and overlay additional series on top of it to see how they relate. This works for individual series or entire groups, so you can compare performance across machines, regions, or time periods.

 

For more complex comparisons, multiple panels with aligned time frames let you view several groups of time series data side by side. The multiple x-axis feature keeps time frames synchronized, so you're always making accurate comparisons—whether you're looking at this quarter versus last quarter or comparing the same period across different years.

 

Customization

Different analysis scenarios call for different visualizations. Time Series Visualization provides flexible Y-axis scaling options: use a global scale for consistent comparison across all series, separate scales so each series renders in its own dynamic range, or adaptive scaling that removes outliers using percentiles to focus on typical behavior.

 

You can assign colors to data series using a color picker or palette, making it easier to maintain visual consistency or highlight specific series that need attention. Zoom in to specific time ranges using keyboard shortcuts or mouse gestures, then use the time slider to move forward or backward through your data without losing context.

 

The crosshair feature displays values for all data series at a specific point in time as you hover over the chart—useful for understanding exactly what was happening at a particular moment. For data with wide value ranges, you can switch between linear and logarithmic scales for both axes.

 

Where this applies

These capabilities support a range of scenarios: IoT and equipment monitoring where you're comparing sensor readings to spot deviations before they become failures; operational analytics where you're tracking response times or error rates across service tiers; and business trend analysis where seasonal patterns and year-over-year comparisons matter.

To learn more, refer to the Real-Time Dashboard specific visuals documentation.

 

Share Copilot Exploration Insights

Collaboration is essential when working with data, and now you can share the insights you discover through Copilot with your team. After using Copilot to explore your data and uncover meaningful patterns, you can generate a shareable link that gives others access to the same query and results you're viewing. You also have the option to include the visual in your shared insights, making it easier for recipients to understand the data briefly.

 

To share your insights, select the share icon in the Copilot pane or in the expanded view. In the share dialog, choose whether to include the visual, and then select Copy link. You can then send this link to colleagues through your preferred communication channel.

 

When recipients open the shared link, they see a view of the query results and the visual, if you include it. From there, they can take several actions with the shared insights:

  • Run or adjust the query to get updated results.
  • Modify the visual type and customize its appearance, if a visual was included.
  • Share the insights again with others.
  • Save the query for an existing or new KQL Queryset for future reference.

 

This capability supports data-driven decision making across your organization by making it simple to distribute findings without requiring recipients to recreate the analysis themselves. Team members can build on shared insights by customizing visuals to suit their needs or saving queries for ongoing monitoring.

 

To learn more about using Copilot for data exploration in Real-Time Intelligence, refer to the Copilot-Assisted Real-Time Data Exploration documentation.

 

Real-Time Dashboards Live Refresh (Generally Available)

Monitoring real-time data effectively has always required a tradeoff: refresh your dashboards frequently to capture every update or refresh less often to reduce compute costs and system load.

 

Live Refresh for Real-Time Dashboards eliminates this compromise by introducing an event-driven refresh model that responds to your data, not a fixed schedule.

 

Instead of polling your data sources at predetermined intervals regardless of whether new information exists, Live Refresh uses lightweight background queries to detect when data is ingested. When new data arrives, the dashboard automatically refreshes the affected visuals. When nothing has changed, no unnecessary queries run. This approach delivers the responsiveness you need for time-sensitive monitoring while significantly reducing compute overhead during quiet periods.

 

For organizations monitoring high-frequency data streams, Live Refresh ensures your dashboards reflect new information the moment it's available—without the cost implications of constant polling. For teams running multiple dashboards across large data volumes, the reduction in unnecessary refresh cycles translates directly to lower operational costs. And when monitoring multiple data sources simultaneously, Live Refresh intelligently refreshes only the visuals with new underlying data rather than refreshing everything on a fixed schedule.

 

The feature also provides flexibility for dashboard viewers. When you need to investigate a specific data point without the visuals changing, you can pause Live Refresh temporarily to analyze the current state. Resume when you're ready to return to real-time monitoring.

 

Live Refresh represents a shift from schedule-based to data-driven dashboard updates. You get fresher data when activity is high and automatic cost savings when activity is low—without manual intervention or complex configuration. Your dashboards stay current when it matters, responding to your data rather than to an arbitrary timer.

 

Eventstream Observability through Workspace Monitoring (Preview)

Ever wonder how much data volume is flowing in and out of your Eventstream? Or wished you could just query the health or errors in one place — without clicking into each Eventstream component? Now you can.

 

Eventstream observability through Workspace Monitoring brings monitoring to your streaming pipelines with a single toggle and zero configuration. Just enable Workspace Monitoring in your workspace settings, and three purpose-built tables are automatically created in your monitoring Eventhouse — no code, minimal setup, and no managing infrastructure.

 

EventStreamNodeStatus tells you briefly the state of nodes in your Eventstream topology — running, paused, or failed. EventStreamMetrics gives you the throughput picture — incoming and outgoing messages, bytes, watermark delay, etc. EventStreamErrorMetrics surfaces exactly what's going wrong: runtime errors, deserialization errors, data conversion errors, etc., all with counts over time.

 

The real power? Everything is queryable in KQL. Build Real-Time dashboards for continuous monitoring in your workspaces. Set up Data Activator alerts to catch throughput drops or error spikes before your customers do. Run trend analysis across hours, days, or weeks to understand how your pipelines behave.

Figure: High-level overview image showing the monitoring Eventhouse with the three Eventstream tables.Figure: High-level overview image showing the monitoring Eventhouse with the three Eventstream tables.

To learn more, refer to the Monitor Eventstream items with Workspace Monitoring documentation.

 

 

What's new in Business Events in Fabric

Business Events in Fabric: Key Capabilities Overview

Business Events in Microsoft Fabric introduce a unified pub-sub model that allows organizations to transform raw signals into meaningful business outcomes and route them across the platform for analytics, automation, and application integration.

The following is an overview of the latest capabilities across publishers, consumers, and platform-level features.

 

Eventstream as a Business Events Publisher (Preview)

Eventstream enables organizations to convert high-volume operational data into meaningful business signals.

Instead of exposing raw telemetry or database changes, Eventstream applies transformations such as filtering, enrichment, and correlation to emit structured Business Events that represent real business moments.

 

Capabilities:

  • Turning low-level changes (e.g., CDC rows) into business events like OrderCreated or HighValueOrderDetected.
  • Centralizing event generation from multiple data sources into a single pipeline.
  • Decoupling internal systems from downstream consumers through a standardized event model.

In practice, Eventstream acts as the signal processing layer of Business Events, where raw data is elevated into actionable business context.

Figure: Eventstream provides a built-in Business Events destination for scenarios where the logic is clearly defined, and the primary goal is to move from signal to action as quickly as possible.Figure: Eventstream provides a built-in Business Events destination for scenarios where the logic is clearly defined, and the primary goal is to move from signal to action as quickly as possible.

 

Activator as a Business Events Publisher (Preview)

Activator can now publish Business Events, turning the conditions it detects into structured, governed signals that your entire organization can discover, consume, analyze, and act on.

 

Whether you’re monitoring KPIs in a Power BI report, watching thresholds in a Real-Time Dashboard, evaluating conditions in a KQL query, or running a Fabric Warehouse SQL query, Activator can publish a business event when a condition is met.

 

This unlocks:

  • Event generation based on real-time detection (e.g., downtime detection, fraud signals).
  • Mapping rule outputs into structured, governed business event schemas.
  • Emitting events that are centrally visible in Real-Time Hub for governance and discovery.

     

Published events are automatically routed to Eventhouse for historical analysis, where teams can query patterns over time, feed AI/ML models, and build operational reports with no additional configuration needed.

 

This positions Business Events as the unified event-driven foundation in Fabric, where data, signals, and decisions come together to drive real-time workflows across the platform.

 

To learn more and walk through an end-to-end setup, refer to the documentation.

 

Analyze Business Events in Eventhouse and Real-Time Dashboards (Preview)

You can now analyze your Business Events in Eventhouse and Real-Time Dashboards. With Eventhouse being enabled by default, every business event your organization publishes is automatically stored and ready to query with KQL.

 


This unlocks:

  • Historical analysis across every business signal from the moment it's published.
  • Real-Time Dashboards that visualize live event patterns, KPIs, and operational activity.
  • Cross-event correlation by joining business signals from different teams and tools in a single queryable store.
  • AI and ML workloads are grounded in a persistent, structured record of business events.

 

Each business event maps to a dedicated KQL table in your Eventhouse database, with no additional pipelines or configuration needed.

 

Vijay_Nagandla_12-1780366104716.png

To learn more, refer to the documentation.

 

 

Business Events Capacity Consumption (Generally Available)

Business Events now follow the same transparent, consumption-based model used by Azure and Fabric events.

 

Capacity consumption is based on two operation types:

  • Event operations (per event): Covers publish, filtering, and delivery. Publish operations are charged to the Event Schema Set item where your business events are defined. Filtering and delivery operations are charged to the consumer’s capacity (e.g., Activator or Eventhouse).
  • Event listener (per hour): Charged to the consumer’s capacity for the duration a consumer is actively listening to your business events.

 

This model gives organizations predictable, scalable consumption that grows with usage, full visibility into costs through the Fabric Capacity Metrics app, and clear separation of charges between publishers and consumers.

 

To learn more, see the capacity consumption documentation.

 

 

Enable Azure Blob Storage Events Without Compromising Your Tenant Private Links Policy

Organizations that enforce tenant-level private links can now receive Azure events (such as Azure Blob Storage events) without relaxing their network security posture. By allowlisting the Azure Event Grid system topic as a trusted resource through Resource Instance Rules, enterprises can maintain strict network boundaries while enabling real-time event workflows from their Azure resources.

 

Previously, enabling Block Public Internet Access at the tenant level blocked all Azure event delivery into Fabric, since these events originate from outside the Fabric tenant. With this improvement, admins can selectively trust specific Azure Event Grid system topics in the workspace inbound networking settings, allowing events from those sources to flow into Fabric while all other public inbound traffic remains blocked.

 

This uses the same Resource Instance Rules that already control inbound access to OneLake, providing a consistent and familiar experience. Fabric events (such as Job events, Workspace item events, and OneLake events) are unaffected by tenant private links since they originate from within the tenant.

 

To learn more, see the documentation.

 

Secure Azure and Fabric Event Flows Across Workspaces with Outbound Access Protection

Azure and Fabric events now support workspace outbound access protection (OAP), enabling enterprises to adopt real-time event-driven workflows across workspaces while maintaining strict network governance. Teams can move faster with confidence that their Activator alerts, Eventstreams, and other event consumers comply with organizational security policies.

 

When OAP is enabled on a workspace, cross-workspace event consumption from that workspace is blocked by default, ensuring consumers can only reach event sources in other workspaces when access is explicitly granted through data connection rules. To allow cross-workspace event consumption, admins add the Real-Time Events connector to the workspace's data connection rules. Event consumption within the same workspace is always allowed, regardless of OAP settings.

 

If OAP is enabled on a workspace that already has cross-workspace event consumers configured, the system detects the change and puts affected configurations in a Paused state, retaining events for up to seven days while the admin resolves the condition.

OAP joins the existing private link support for Azure and Fabric events, giving you multiple layers of network security for your event-driven workflows.

 

To learn more, see the full announcement blog post and the documentation.

 

Secure the data transmission between your MQTT broker and Eventstream connector (Preview)

MQTT is a lightweight, publish-subscribe messaging protocol widely adopted in IoT scenarios. For organizations streaming IoT data from MQTT brokers — whether in smart factories, connected vehicles, or energy systems — securing the connection between your broker and the cloud is critical. Compliance standards often mandate mutual TLS (mTLS) authentication, and many production MQTT brokers only accept clients that present trusted certificates.

 

In April, we introduced Custom CA and mTLS support for Kafka-based sources and Confluent Schema registry in Eventstream, allowing customers to bring their own certificates from Azure Key Vault to securely connect to sources that require custom CA or mutual TLS authentication. We're now extending this capability to the MQTT source connector in preview. You can specify your own CA and client certificates stored in Azure Key Vault, and Eventstream will use them to establish a mutually authenticated, encrypted connection with your MQTT broker — enabling secure, compliant real-time data ingestion from IoT devices operating across untrusted networks.

 

Figure: Custom CA and mTLS support in MQTT source connector.Figure: Custom CA and mTLS support in MQTT source connector.

 

To learn more about the configuration, refer to the MQTT source connector configuration page.

 

Extend IoThub source Eventstream connector with new capabilities (Preview)

Azure IoT Hub is a central message broker for IoT solutions, enriching every device-to-cloud message with valuable system metadata — such as iothub-connection-device-id (the originating device), iothub-enqueuedtime (when the message was ingested), iothub-message-source (whether it's telemetry, a twin change, or a lifecycle event), and more.

 

This metadata is essential for real-world IoT scenarios: calculating end-to-end latency, filtering telemetry by device, auditing message provenance, and building conditional processing pipelines. However, the current IoT Hub source connector in Eventstream only ingests the event payload — it does not retain these system or user metadata properties, limiting downstream analytics and processing.

 

We're now releasing an enhanced IoT Hub source connector built on the Kafka Connect framework (preview). This new connector preserves all event metadata — both system properties and user-defined application properties — by copying them into the user metadata section when data lands in Eventstream.

 

System metadata keys are prefixed with ‘___src__’ to distinguish them from user-defined properties. For example, the IoT Hub system property ‘iothub-connection-device-id’ becomes ‘___src__iothub-connection-device-id’ in the event metadata within Eventstream. To use this enhanced connector, simply select Extended features when configuring your Azure IoT Hub source in Eventstream.

 

Figure: Azure IoT Hub extended connector.Figure: Azure IoT Hub extended connector.

 

 Once your events are flowing with full metadata, you can access these properties using the built-in GETMETADATAPROPERTYVALUE function in Eventstream's SQL operator. For example, to extract all user properties (including the copied system metadata) into a derived stream:

 

SELECT *,

GETMETADATAPROPERTYVALUE ( [your-iothub-stream], '[User]' )

AS UserProperties

INTO [DerivedStream]

FROM [your-iothub-stream]

 

This enables powerful downstream scenarios — such as joining device identity with telemetry for per-device dashboards, computing ingestion latency by comparing ___src__iothub-enqueuedtime against processing time, or routing events based on ___src__iothub-connection-device-id to separate streams for telemetry versus device lifecycle events.

 

To learn more about this connector, refer to the Azure IoT Hub extended connector capabilities.

 

Eventstream streaming connectors for Apache Kafka and Azure Service Bus (Generally Available)

Apache Kafka has become the backbone of real-time data streaming across the industry — trusted by more than 80% of Fortune 100 companies and thousands of organizations worldwide to power mission-critical pipelines spanning fraud detection, IoT telemetry, application monitoring, and AI-driven analytics. Azure Service Bus, as Microsoft's enterprise messaging backbone, is widely adopted across industries including financial services, healthcare, and e-commerce for reliable, decoupled communication between distributed applications and microservices. As organizations increasingly rely on these platforms for real-time workloads, they need seamless, secure, and production-ready ways to bring these streams into their analytics platforms.

 

 

Fabric Eventstream, part of Real-Time Intelligence, provides a no-code experience for ingesting real-time data from a wide range of streaming sources and routing it to multiple destinations for analytics. At the heart of this experience are Eventstream's streaming connectors, which handle the heavy lifting of connecting to external data sources, managing authentication, and ensuring reliable data delivery.

 

The Apache Kafka connector release brings key enterprise-readiness enhancements that strengthen security and broaden compatibility with production Kafka deployments:

  • SASL_SSL security protocol support: In addition to the existing SSL protocol, the Apache Kafka connector now supports the SASL_SSL security protocol — the most widely adopted authentication mechanism in production Kafka environments. SASL_SSL combines SASL-based authentication (such as PLAIN or SCRAM) with SSL/TLS encryption, enabling secure, authenticated connections to Kafka clusters that require both identity verification and data-in-transit encryption.
  • Custom CA and mTLS support: For organizations whose Kafka clusters use certificates issued by private or internal Certificate Authorities, or require mutual TLS (mTLS) authentication, the connector now allows you to bring your own CA and client certificates stored in Azure Key Vault. Eventstream will use these certificates to establish a mutually authenticated, encrypted connection with your Kafka broker — enabling secure data ingestion from Kafka environments with strict certificate requirements.

     

The Azure Service Bus connector has also reached general availability, delivering production-grade stability and performance for enterprise messaging workloads. Whether you're processing order transactions, orchestrating event-driven microservices, or ingesting workflow events, the Service Bus connector provides reliable, low-latency streaming from your Service Bus queues and topics directly into Fabric for real-time analytics.

 

Now, these Eventstream's streaming connectors are production-ready for the most demanding enterprise scenarios — whether you're connecting to self-managed Kafka clusters, cloud-hosted brokers, Azure Service Bus namespaces, or hybrid deployments with stringent security and reliability requirements.

 

Pagination support for Eventstream HTTP connector (Preview)

The Eventstream HTTP connector now supports pagination, helping you ingest data from REST APIs that return results across multiple pages. This update supports both page-based and cursor-based pagination methods.

 

Many APIs limit the number of records returned in a single response. Before this update, working with paginated APIs often required custom code or extra orchestration to request each page and combine the results. With pagination support in the Eventstream HTTP connector, you can continuously request data across pages based on the pagination settings you configure.

Figure: Screenshot of configuring Pagination method for Eventstream HTTP Connector.Figure: Screenshot of configuring Pagination method for Eventstream HTTP Connector.

 

Pagination support includes:

  •  Page-based pagination for APIs that use page numbers, page size, offsets, or similar request parameters.
  •  Cursor-based pagination for APIs that return a cursor, continuation token, or next-page marker in the response.
  • Configuration options that let Eventstream continue requesting pages until the API indicates that no more data is available.

 

This capability is useful when you ingest activity logs, audit records, operational events, or other API data from services that expose paginated HTTP endpoints. You can use the HTTP connector to bring the data into Eventstream, apply transformations, and send the results to destinations such as Eventhouse, Lakehouse, or Activator.

 

To learn more about Eventstream HTTP Connector and its pagination capability, refer to the HTTP Source in Fabric Eventstream documentation.

 

Databases Apps

Fabric Apps: a new way to build and run application backends with the Rayfin SDK (Preview)

Rayfin is an open-sourced SDK and CLI that provides everything you need to power the backend for modern applications. When deployed to Fabric, the Fabric app comes with an enterprise-grade backend, including a built-in database, authentication, and front-end hosting, all integrated with your existing Fabric data sources, starting with Semantic Models.

 

Building with an agent? Describe what you want, and your agent can build it. With the Rayfin SDK, your data schema, backend services, and access policies are all defined in code, so the agent has the context it needs to generate your app logic and wire up your backend. Connecting to existing Fabric data sources is just as straightforward, and your current access policies are respected automatically.

 

Now you can build that inventory tracker, revenue dashboard, or whatever app you've been dreaming of, and have it governed like any other Fabric artifact. Manage your app, set user permissions, and configure underlying resources like the database right from your Fabric workspace, and any new data lands right in Fabric, ready for BI and AI workloads. We can't wait to see what you build.

 

Picture1.gif

To learn more, refer to the Fabric Apps documentation.

 

Data Factory

CDC with SQL estates in Copy job (Generally Available)

Change Data Capture (CDC) in Copy job for SQL-based sources including Azure SQL Database, SQL Server and Azure SQL MI, bringing enterprise-grade, production-ready incremental data movement to Fabric.

 

With native CDC support, Copy job can automatically capture and replicate inserts, updates, and deletes directly from supported source databases, keeping your destination continuously in sync without complex pipeline logic. This eliminates the need for manual change tracking or custom incremental logic. Instead, Copy job intelligently detects CDC-enabled tables and applies the exact changes downstream, ensuring high fidelity replication with minimal overhead.

 

To learn more, refer to the Change data capture (CDC) in Copy Job documentation.

 

Extended SCD Type 2 Support in Copy job for Fabric Warehouse (Preview)

Native SCD Type 2 supported in Copy job helps preserve full change history with effective dating and built‑in soft delete handling. This industry‑standard pattern allows you to track how each record evolves over time while ensuring deletes are handled gracefully.

This capability is now extended to Fabric Warehouse, giving you broader coverage with the same consistent SCD Type 2 experience across sources and destinations.

 

With native SCD Type 2 in Copy job, you can automatically maintain historical versions of your data and capture deletes as soft deletes—meaning when a record is removed at the source, the corresponding row in the destination is marked as inactive instead of being physically deleted. The result is a complete, audit‑ready view of your data’s lifecycle, including historical records that no longer exist in the source system.

 

Figure: Enabling SCD2 in Copy job.Figure: Enabling SCD2 in Copy job.

 To learn more, refer to the Change data capture (CDC) in Copy Job documentation.

 

Extended Auto partition Support in Copy job for Oracle, Fabric Lakehouse and SAP Hana (Preview)

Auto Partitioning (Preview) in Copy job delivers significantly higher performance by automatically parallelizing data movement based on source characteristics—without requiring manual tuning. Now, we’re taking this further by extending Auto Partitioning support to more connectors, including Oracle, Fabric Lakehouse, and SAP HANA.

 

With this expansion, Copy job can automatically scale out read and write operations across these data stores, maximizing throughput with simplified experience. You no longer need to predefine partition columns or manage custom sharding logic—Copy job intelligently determines the optimal partitioning strategy for you.

 

To learn more, refer to the What is Copy job in Data Factory documentation.

 

Edit Copy Job via JSON Payloads for Maximum Flexibility

Copy jobs introduce the ability to edit directly through JSON payloads, giving advanced users maximum flexibility and precision when configuring your Copy job for data movement. This capability provides full flexibility in Copy job authoring, beyond what is available through the visual authoring experience.

 

With JSON-based editing, you can programmatically define and update Copy job configurations, making it easy to automate job edit and apply bulk changes. This is especially powerful for large-scale edit or scenarios where consistency and repeatability across environments are critical.

Figure: Edit Copy job via JSON Payloads.Figure: Edit Copy job via JSON Payloads.

To learn more, refer to the What is Copy job in Data Factory documentation.

 

 

Integrate SAP data using Copy job with ABAP Add-On (Preview)

Copy job for SAP via ABAP Add-On extends the capabilities of Fabric’s built-in SAP connectivity (SAP HANA, SAP Table, and SAP BW OpenHub). Instead of relying only on standard SAP APIs, this new option provides a plug-in (aka ABAP Add-On) that customer can install on the SAP server, enabling more advanced data extraction logic and helping address some of the limitations of existing APIs.

 

Capabilities:

  • Effortlessly move SAP data in bulk or incrementally to any destination across clouds with flexible table and column mapping.
  • Use the Microsoft-built ABAP Add-On to achieve advanced SAP data ingestion for complex scenarios.
  • Combine SAP data with other sources for holistic analytics and AI, orchestrating end-to-end data movement and transformations in a single pipeline.

Figure: Use SAP ABAP Add-On to copy data from SAP systems in Copy Job.Figure: Use SAP ABAP Add-On to copy data from SAP systems in Copy Job.

 

To learn more and get started, refer to the Copy Job for SAP via ABAP Add-On documentation.

 

 

Business workflow management in Fabric Pipelines with a net-new Approval activity (Preview)

Business Workflow Management capabilities are new in Fabric pipelines.

 

This is a meaningful expansion of what pipelines represent. Rather than being limited to data movement and transformation, pipelines can now act as the backbone for end-to-end operational workflow coordinating systems, processes, and people together.

 

With these capabilities, you can:

  • Combine automated data workflows with human approvals and decision points
  • Model real-world business processes (not just data pipelines)
  • Maintain visibility and control across both system-driven and human-driven steps

This brings Fabric Data Factory closer to a true orchestration platform for business processes, not just data engineering—unlocking entirely new scenarios that were previously difficult or fragmented to implement.

One of the most exciting additions this month is the introduction of the Approval activity, bringing human decision-making directly into your pipelines.

Figure: Setting up the Approval activity settings with an Outlook 365 Email type approval request.Figure: Setting up the Approval activity settings with an Outlook 365 Email type approval request.

 

 

Figure: The approver experience in the Approval activity flow.Figure: The approver experience in the Approval activity flow.

 

 

Previously, pipelines have primarily automated system-driven tasks. With Approval activity, you can now incorporate human-in-the-loop workflows and API-driven approvals, where pipeline execution pauses and waits for an approval or decision before continuing.

 

This unlocks scenarios such as:

  • Business approvals before promoting data or models to production.
  • Validation checkpoints for sensitive operations.
  • Exception handling workflows that require manual intervention.

By embedding approvals directly into pipelines, you no longer need to stitch together external systems or manual coordination steps. Instead, your workflows remain centralized, auditable, and fully orchestrated in one place.

 

To learn more, refer to the Approval activity documentation.

 

 

New Refresh Materialized Lake View activity in Fabric Pipelines (Preview)

Keeping downstream analytics up to date is critical—but managing refresh logic across systems can quickly become complex and costly.

 

With the new Refresh Materialized Lake View activity, you can now trigger and manage refresh operations directly inside your pipelines, ensuring that transformations and consumption layers stay in sync with upstream data.

Figure: The Refresh Materialized Lake View activity and its easy-to-set-up settings.Figure: The Refresh Materialized Lake View activity and its easy-to-set-up settings.

What makes this especially powerful is Fabric’s built-in optimization strategy. Each time a refresh runs, Fabric automatically determines whether a full refresh, incremental refresh, or no refresh is needed based on changes in the underlying data. This helps reduce unnecessary compute usage while still delivering fresh, reliable data for downstream consumers.

 

By bringing this capability into pipelines:

  • Orchestrate refreshes as part of your end-to-end data workflows.
  • Align refresh timing with upstream ingestion and transformation steps.

 

Reduce operational overhead by relying on intelligent refresh decisions. This is another step toward making Lakehouse operations fully integrated and automated within your pipeline experience.

 

To learn more, refer to the Refresh Materialized Lake View activity documentation.

 

Refresh SQL Endpoint activity in Fabric Pipelines (Generally Available)

The Refresh SQL Endpoint activity now provides a reliable and fully supported way to keep your SQL analytics layer up to date as part of your pipeline workflows.

Figure: Refresh SQL Endpoint activity and its settings.Figure: Refresh SQL Endpoint activity and its settings.

 

You can seamlessly incorporate SQL refreshes into your orchestration logic, ensuring that downstream reporting and analytics always reflect the latest data—without requiring separate processes or tools.

 

To learn more, refer to the Refresh SQL Endpoint activity documentation.

 

Modern Fabric Pipeline node experience now available (Preview)

As pipelines grow and complexity, usability becomes just as important as capability.

 

This month, we’re introducing a new Pipeline node experience designed to make working with large, complex pipelines significantly easier. The updated canvas improves how you navigate, visualize, and manage your workflows—especially when dealing with deeply nested or high-scale pipeline graphs.

Figure: A look at the new Pipeline node UI.Figure: A look at the new Pipeline node UI.

 

 

With this new experience, you’ll notice:

  • Improved visibility when zooming in and out of large pipelines.
  • A cleaner, more structured layout for complex orchestration logic.

 

A more responsive and scalable canvas experience. These changes are designed specifically to support modern, enterprise-scale pipelines, where workflows often span dozens or even hundreds of activities.

 

The new experience is available as an opt-in preview, so you can start exploring it and provide feedback as we continue to refine the design.

 

To learn more, refer to the new Pipeline node experience documentation.

 

 

Conditional activity retries in Data Factory pipelines (Preview)

When a pipeline activity fails, you shouldn't have to choose between retrying everything or nothing. Conditional activity retries give you fine-grained control over retry logic by letting you define exactly when an activity should retry based on the error it returns.

 

Set retry conditions using error code, error message, or both — with AND/OR logic — so your pipelines can automatically recover from transient failures while fast-failing on errors that won't self-resolve. No custom error handling code, no wrapper activities, no unnecessary retries burning your compute budget.

Vijay_Nagandla_0-1780365282987.png

 

This is especially valuable for long-running pipelines that call external APIs, run Databricks jobs, or execute stored procedures where certain error classes are retriable and others are not.

 

Connection and Item Reference Support for Variable Libraries in Pipelines

We’re introducing connection and item reference support in Variable Libraries for Fabric Data Factory pipelines, enabling pipelines to dynamically bind to external connections and Fabric resources without hardcoding configuration values.

 

With this update, pipelines can use two new Variable Library variable types:

  • Connection references to securely point to external data sources without embedding connection details.
  • Item references to dynamically link to Fabric resources such as Lakehouses, notebooks, or pipelines.

These variables integrate directly into pipeline authoring through dynamic content, allowing activity settings to resolve connections and resources at runtime.

 

This update simplifies pipeline development and deployment by centralizing configuration in Variable Libraries. Pipelines can now run across environments using the same definition, with connections and resource references resolved based on the active configuration, reducing manual updates and configuration drift, and enabling more reusable, maintainable, and environment-aware pipelines for enterprise-scale data integration.

 

Learn more about Variable library integration with pipelines.

 

 

Variable Library support in Apache airflow job

Managing environment-specific configuration across dev, test, and production Airflow workloads just got simpler. Fabric Airflow jobs now support Variable Libraries — a workspace-level store of key-value configuration that your DAGs can reference at runtime.

Instead of hard-coding values in DAG code or duplicating Airflow variables across environments, you define a single Variable Library with multiple value sets (for example, dev, test, prod), mark one active per workspace, and Airflow resolves the right values automatically. No DAG edits, no redeployment — just switch the active value set as you promote across workspaces.

 

This brings Airflow in line with Fabric's CI/CD patterns and removes a common onboarding blocker for teams migrating from ADF or self-hosted Airflow, helping you build more maintainable and scalable orchestration workflows.

 

Workspace Identity authentication for SharePoint

As part of our continued investment in secure, enterprise‑ready connectivity, we’re rolling out Workspace Identity authentication support for SharePoint in Microsoft Fabric.

 

This update helps customers transition away from legacy authentication models as Azure ACS is retired, while enabling more secure, service‑to‑service access patterns.

 

With this release, supported Fabric experiences can authenticate to SharePoint using workspace identity, allowing the Fabric workspace itself to securely access SharePoint resources—without relying on user credentials or legacy ACS‑based authentication.

 

This support is rolling out for:

  • Power BI Service
  • Fabric Dataflows Gen2

Many organizations rely on SharePoint as a source for operational and business data. Historically, automating access to SharePoint requires user‑based credentials or legacy service models that don’t meet modern security standards.

 

Workspace identity provides:

  • Secret-less, managed authentication
  • Better alignment with Microsoft Entra ID governance
  • A scalable path forward as ACS is retired

     

Workspace Identity support for SharePoint applies to supported Fabric‑hosted experiences. Availability may vary by workload and host, and customers should refer to updated documentation for supported scenarios and configuration guidance.

We’ll continue to expand workspace identity support across additional connectors and workloads, and we appreciate the feedback that’s helping shape these investments.

 

Learn more about workspace identity authentication in Microsoft Fabric on Microsoft Learn.

 

Snowflake now supports Secret-less Authentication using Microsoft Fabric Workspace Identity

We’re continuing to strengthen enterprise‑grade connectivity across the Power BI and Microsoft Fabric ecosystem. The Snowflake connector in Power Query now supports secret-less authentication using Microsoft Fabric workspace identity, enabling secure, identity‑based access to Snowflake data without storing usernames, passwords, or long‑lived secrets.

With this update, the Snowflake connector in Power Query supports:

  • Secret-less authentication using Microsoft Entra ID backed by Fabric workspace identity
  • Usage in Microsoft Fabric–hosted Power Query experiences, including Power BI semantic models and Fabric Dataflows Gen2

A clear path away from legacy, credential‑based authentication models.

 

Note: Username/password authentication for Snowflake is being deprecated by Snowflake. Microsoft Entra ID–based authentication is the recommended approach moving forward.

 

Learn more in the updated Power Query Snowflake connector documentation.

 

Google BigQuery V2 Connector

As customers scale analytics across cloud data platforms, reliable, high‑performance connectivity is foundational.

The Google BigQuery connector V2 for Power Query is a modernized connector designed to deliver better performance and a stronger foundation for analytics workloads in Microsoft Fabric.

 

The original connector was built on ODBC, which works well for many scenarios but can become limiting as data volumes and concurrency increase. V2 is built on Apache Arrow Database Connectivity (ADBC), a cloud‑native standard optimized for high‑throughput analytical data transfer. This enables more predictable performance, improved reliability, and modern architecture aligned with the long‑term direction of Power Query and Fabric.

 

With the V2 connector, customers can access Google BigQuery using an ADBC‑based implementation across Power BI, Dataflows Gen2, and other Fabric‑hosted Power Query experiences. All future enhancements to the Google BigQuery connector will be delivered through V2.

 

In Microsoft Fabric, the V2 connector integrates directly with Power Query–based experiences like Dataflows Gen2, making it easier to bring BigQuery data into OneLake‑powered analytics workflows. For new and modernized Fabric workloads, V2 is the recommended option.

 

You can start using the Google BigQuery connector V2 today in supported Power Query experiences. Refer to Microsoft Learn for setup instructions and detailed capability information.

 

Mirroring in Fabric advancing network security controls

We’re expanding network security support for mirroring in Fabric. Fabric provides multiple layers of inbound network protection to help organizations control where connections can originate when accessing data and items.

 

For workspaces that restrict public access and only allow connections from selected networks and private links, you can now set up mirroring for Azure SQL Database, SAP, SQL Server (2016–2022), and SharePoint. This expands the existing support for open mirroring and mirroring for Azure Cosmos DB, Azure SQL Managed Instance, and SQL Server 2025.

 

To learn more, refer to the Mirroring network security documentation.

 

 

Until next month

That’s a wrap for June!

As always, we’re continuing to build based on your feedback—especially in areas like data agents, real-time scenarios, and enterprise-scale management. Many of the features in this update are still in preview, so if you’re trying them out, your input is especially valuable in shaping what comes next.

For more details, documentation, and deep dives on any of these features, check out the links throughout this post—and keep the feedback coming.

 

Comments