Fabric Weekly
Issue 2
7th July 2026
🎉 Welcome to issue #2 of Fabric Weekly!
On the official side: the headline is the Fabric data agent API going public - the data agent SDK now runs on the Fabric REST API instead of being stuck inside a notebook, so you can create, configure, and publish data agents from your own CI/CD pipelines, internal tools, or backend services. Big news for anyone provisioning the same agent shape across dozens of workspaces.
Also worth a look: Delegated OneLake Shortcuts hit preview, letting a shortcut route access through a service principal instead of every consumer's own identity- This solves both the "central team can't scale permissions" problem and cross-tenant sharing. And Item Recovery is now GA, so "I accidentally deleted the wrong lakehouse" is a recoverable mistake!
Over in the community: Reza Rad shows how pairing GitHub Copilot with Microsoft's open-source Fabric Skills library turns "list every empty workspace in my tenant" from a manual job into a single prompt.
Bart Wullems shows how Fabric's Copy job now builds Slowly Changing Dimension Type 2 for you with zero merge statements - provided your source/destination pair is currently supported. And on the reporting side, Marco Russo and Alberto Ferrari debug a running-total measure that quietly broke because of sort-by-column settings, fixing it properly by packaging REMOVEFILTERS into a reusable user-defined function instead of patching the one measure.
Last but not least, Kevin Chant shows how Managed DevOps Pools can get your Fabric deployment pipelines into a firewall-protected Key Vault without the usual self-hosted-agent headache.
🤖 Copilot, AI & Agents
- Fabric data agent API is now public: Build Fabric data agents into your tools and pipelines The Fabric data agent SDK can now run outside Fabric entirely, via the newly public REST API — so teams can create, configure, and publish data agents from their own CI/CD pipelines, internal tools, or backend services instead of being limited to the Fabric notebook environment. Useful for ISVs and anyone provisioning the same data agent shape across many workspaces.
📊 Reporting & Insights
- Supercharge your real-time data ingestion: What's new in Fabric Eventstream connectors A batch of Eventstream connector updates hits GA — private network support, an Apache Kafka connector, Azure Service Bus connector, and custom CA/mTLS auth — plus new previews for workspace-identity auth on Event Hubs, IoT Hub metadata, Oracle CDC, and HTTP pagination.
- Understanding the TREATAS Function in DAX with a Real Use Case TREATAS creates a "virtual relationship" so Power BI can apply filters from one table onto an unrelated one — demonstrated by having a disconnected date-table slicer filter a separate sales table. Useful for disconnected slicers, what-if parameter tables, and Budget-vs-Actual comparisons, but the piece cautions that physical relationships should still be the first choice, since virtual relationships can slow query performance on large datasets.
- Using REMOVEFILTERS in DAX user defined functions Marco Russo and Alberto Ferrari debug a real problem: a running-total measure was returning wrong results because sort-by-column settings on weekday fields kept filters active that time-intelligence functions expect cleared. The fix packages the needed REMOVEFILTERS calls into a reusable DAX user-defined function (Gregorian.RemoveFilterKeepColumns), keeping the fix consistent as the calendar table gains more filter-keep columns.
⚙️ Data Engineering
- AI-assisted Synapse Spark and pipeline migration to Microsoft Fabric from the command line (Preview) New open-source "skills" automate Synapse-to-Fabric migration for Spark pools, notebooks, and Data Factory pipelines from the command line — either a quick lift-and-shift or a guided migrate-and-modernize path with automatic code refactoring. Works with GitHub Copilot, Claude, and other AI coding tools; flags unsupported items like GPU pools upfront.
- Rethinking Dimensional Modeling: Building a Fabric Data Warehouse with Materialized Views Argues that Fabric's Materialized Views let architects keep the traditional star schema while cutting the ETL overhead of hand-built aggregate tables — precomputing things like monthly sales by region, complex fact-dimension joins, and historical snapshots so Power BI dashboards query results instantly instead of recalculating them each time.
- Optimizing ADF Copy Activity Performance: DIU Tuning, Parallelism, and Staging A deep dive into tuning Azure Data Factory Copy Activity — the same engine behind Fabric Data Factory pipelines. Covers Data Integration Units (DIUs, ranging 4–256), parallel copy settings, and staged copy through Blob/ADLS, plus where to look in the monitoring view (queue time, transfer time, listing time) to diagnose bottlenecks.
- Slowly Changing Dimensions – An introduction A primer on Slowly Changing Dimensions covering the full taxonomy — Type 0 (never changes), Type 1 (overwrite), Type 2 (keep full history via surrogate keys), Type 3 (keep just the previous value), and Type 4 (split current/history tables) — with guidance on picking the right type per column. Sets up Bart's companion post on implementing Type 2 natively in Fabric's Copy job.
- Slowly Changing Dimensions in Microsoft Fabric - The no-code way Fabric's Copy job now offers SCD Type 2 as a built-in write method — it auto-adds Valid_From/Valid_To/Is_Current columns via CDC, no merge logic required. But it's preview-stage: only works with select destinations (Azure SQL, Fabric Lakehouse, SQL DB in Fabric — not yet Snowflake/Oracle sources or Fabric Data Warehouse), requires CDC on the source, needs Copy job to create the destination table itself, and breaks if you edit column mappings after enabling it.
🏛️ Storage & Platform
- Simplifying secure data access with Delegated OneLake Shortcuts (Preview) OneLake Shortcuts get an optional second auth mode: instead of every user hitting the target with their own identity, a Delegated Shortcut routes access through a configured identity (service principal, org account, or workspace identity) — letting a central team scope what that identity sees while business units manage their own end users. Also enables cross-tenant sharing (e.g. test-to-production) with zero data copies.
- Securing zero-copy distribution patterns with OneLake security and shortcuts A companion deep-dive to the Delegated Shortcuts announcement: explains exactly how OneLake security combines with passthrough vs. delegated shortcuts (permissions intersect with a delegated identity's own access — never exceed it), and gives concrete guidance on when each model fits, e.g. passthrough for internal collaborative engineering, delegated for governed publishing to large or external audiences.
- Spread your SQL Server wings with Microsoft Fabric in 2026 Kevin Chant's 2026 refresh of his "spreading your SQL Server wings" series — a holistic tour of Fabric's eight workloads built on OneLake, aimed at SQL Server professionals (and anyone eyeing a Fabric architect role who needs the full picture). Maps specific SQL Server skills onto Fabric equivalents: Power BI modeling, T-SQL in the Data Warehouse's query editor, SSIS know-how transferring to Data Factory pipelines, and existing Azure DevOps/GitHub CI/CD experience extending to Fabric Git integration and fabric-cicd.
- Azure Data Lake Storage (ADLS Gen2) to Microsoft Fabric Migration A practical comparison of three ways to move ADLS Gen2 data into Fabric: Copy Job (simple one-time physical copy, but duplicates storage), OneLake Shortcuts (no data movement, lowest cost, but depends on the source staying available), and Data Pipelines (production-grade with scheduling and incremental loads, more setup). Recommends a hybrid approach — Copy Job for historical data, Pipelines for ongoing loads, Shortcuts for direct access.
- SaaS-ifying the Data Estate Using Microsoft Fabric Argues Fabric turns the data estate into a SaaS-like platform: OneLake removes storage silos, Direct Lake Mode lets Power BI query it without importing/duplicating data, and built-in RLS/sensitivity labels keep Copilot-generated insights inside existing governance boundaries. Also points to the rise of the "Analytics Engineer" — a hybrid role spanning data engineering and BI now that Fabric collapses those tools into one platform.
🔐 Governance & Security
- Workspace Outbound Access Protection (OAP) for Real-Time Intelligence (Preview) Outbound Access Protection now covers Real-Time Intelligence workspaces, blocking outbound connections by default. Most KQL analytics work is unaffected since it queries data already in Fabric — the one real restriction in this preview is that Copilot-assisted KQL query authoring won't work in a protected workspace, while manual query authoring and existing investigation workflows continue as normal.
💰 Management & Cost
- Item Recovery in Microsoft Fabric (Generally Available) Item Recovery is now GA: recover accidentally deleted Fabric items across supported workloads via the Workspace Recycle bin or REST API, with tenant-wide retention policies admins can configure.
- No Code, No Tools: Documentation for Power BI and Fabric Using Fabric Skills and GitHub Copilot Reza Rad shows how pairing GitHub Copilot with Microsoft's open-source Fabric Skills library (github.com/microsoft/skills-for-fabric) lets you inventory workspaces, audit stale Power BI reports, and generate full semantic-model and report documentation from plain-English prompts — no scripts or portal clicking required.
- Use Managed DevOps Pools for Microsoft Fabric deployments Kevin Chant shows how Managed DevOps Pools can drive Microsoft Fabric deployments, letting pipelines securely reach firewall-protected Azure Key Vaults without the complexity of self-hosted agents. Builds on his earlier post on the same setup.