What’s new across Microsoft SQL at SQLCon/FabCon Europe 2026

Live From Barcelona

Woo! I’m posting this live from the first ever European SQLCon (joint conference with FabCon EU), and the second ever SQLCon. The energy here is great, as the community comes together in Barcelona and as we prepare to go on the keynote stage and share what’s new! A heartfelt nod to the internal team, our customers, and the community who all contributed to what we are releasing today – the feedback to build to feedback loop is getting tighter, and each is critical for Microsoft SQL to build a better platform.

If you’re here in Barcelona, the full schedule from the Microsoft SQL product team is at aka.ms/sqlsessions. We’d love to chat about your scenarios, so come find us in sessions and at the booth.

And if you’re reading this from home, it’s likely post-keynote, but that doesn’t mean you will miss out on the news! Here’s what’s new across Microsoft SQL: SQL Server, Azure SQL, SQL database in Fabric, and the tools you use every day.

Hybrid and migration

SQL Server on Azure Local is generally available in both connected mode and disconnected mode. Disconnected mode matters if you operate in air-gapped, sovereign, or intermittently connected environments and still want the Azure management surface for your SQL estate.

Azure Arc-enabled SQL Server expands to Germany West Central and Italy North, continuing the regional buildout so you can keep data residency commitments while running Arc management.

SQL Server on Azure Virtual Machines in Azure Bleu (France Sovereign Cloud) is generally available, bringing the SQL VM experience to France’s sovereign cloud.

For VM provisioning specifically, you can now configure ephemeral storage during provisioning on v6+ VMs. If you’ve been resizing disks post-deployment to get tempdb where you want it, that step goes away.

On the Linux side, two things landed, the bulkadmin role is now enabled for SQL Server on Linux so bulk inserts work the way they do on Windows, and a script-based provisioning experience (preview) enables flexible and customizable SQL Server deployment experience for Linux Azure VMs.

Migration got a real upgrade this cycle:

• Skills for SQL migration to Azure (preview) brings assessment, migration, and validation into an agent-driven workflow instead of a checklist you drive by hand.
• Migrations to Azure SQL Database Hyperscale from the Arc portal (uses Azure Database Migration Service and SHIR) are in preview.
• SSMS target sizing and pricing recommendations are generally available: configuration-based Azure SQL target recommendations with savings options, so the “what will this cost me” question gets answered inside the tool doing the assessment.
• Expert migration assistance in SSMS connects you to help when the automated path isn’t enough.

Enterprise-ready SQL estate

The Fabric Database Hub estate view (now in preview) along with Database Agents provide a centralized control plane for monitoring and managing your entire database estate across SQL Server, Azure SQL, PostgreSQL, Azure Cosmos DB, and SQL databases in Fabric. Its unified estate view brings inventory, health, and risk signals together in a severity-prioritized experience, so administrators can quickly see what needs attention instead of managing resources one blade at a time. Findings are organized across Security, Performance, and Optimization, allowing you to move directly from an estate-level summary to the affected resource and issue details.

The Azure portal SQL query editor is generally available. Run queries against Azure SQL Database from the portal without a client install, without a jump box, and without leaving the resource you were already looking at. We improved this experience with our learnings from Fabric, and it’s [finally] generally available. In the Azure portal, we also turned Zone redundancy on by default for Hyperscale, General Purpose, Business Critical, and Premium. Defaults shape outcomes, and we want the resilient choice to be one you get without thinking about it.

Hyperscale keeps expanding with new options for both high-performance and cost-optimized workloads:

• 160 and 192 vCore Premium-series service objectives are generally available for the most compute-demanding applications.
• Hyperscale serverless auto-pause and auto-resume (preview) helps customers reduce costs for development, test, and other active workloads.
• Hyperscale elastic pools expanded capacity (preview) raises the ceiling to 50 databases and 128 TB per pool.

Automatic index compaction is generally available. Index bloat gets handled by the platform instead of by a maintenance window and a script. But seriously, some are saying this may be the end of index maintenance, don’t sleep on this one!

SQL Managed Instance picked up a set of updates that mostly fall under “the platform does more so you configure less”: zone-redundant next-gen General Purpose (GP) and flexible memory in next-gen GP (preview), flexible memory for Business Critical (GA), higher log rate for Business Critical (GA), support for multiple databases in Always On Availability Groups for MI link (preview), and SQL performance monitoring (preview).

Local time zone support is in public preview for Azure SQL Database and Azure SQL Managed Instance (Always-Up-To-Date policy). Set the database or session time zone once and stop writing UTC time zone conversion math into every query or expression.

Immutability for point-in-time-restore backups is generally available for Azure SQL Database General Purpose and Business Critical tiers, so your backup chain can meet retention and tamper-resistance requirements directly. You don’t have to configure it and you don’t have to pay extra for it, you just get it.

SQL database in Fabric gets workspace-level Private Link (preview) for network isolation at the workspace boundary, and SQL Server migration to SQL database in Fabric is generally available. We also shipped several editor improvements that hopefully improve your quality of life (yes, you can finally resize the column widths in the results!). Last month, we also released integrations and support for Skills for Fabric.

Vector indexes are generally available (Azure SQL DB/MI, Fabric SQL). DiskANN index on vector columns, in the same database as your operational data, with the same security model and the same T-SQL. No separate vector store to provision, secure, and keep in sync.

AI and developer experiences

The tooling story this cycle is about agents doing multi-step work while the deployment discipline underneath gets stronger.

GitHub Copilot Agent Mode in SSMS is generally available. Agent Mode reads context across your database objects, reasons through multi-step problems, and executes workflows: trace a dependency chain, evaluate an execution plan, generate the DDL, show you the diff. It’s now integrated into Solution Explorer, and you can export chat so the reasoning behind a change survives past the session.

SQL Projects in SSMS are generally available, and so is Schema Compare. Together they close the loop between what an agent generated and what actually ships: compare, diff, generate the migration script, commit it. Your schema belongs in source control, and now the path to get it there is a first-class part of the tool.

SQL Formatter is generally available in both SSMS and the MSSQL extension for VS Code. Consistent formatting at keystroke speed matters more when a meaningful share of your T-SQL is machine-generated.

Data API builder is adding vector and JSON data type support across releases 2.1, 2.2, and 2.3, so more of your existing schema maps cleanly to generated REST, GraphQL, and MCP endpoints.

Mirosoft SQL Agent skills are in preview: packaged, reusable skills that give AI agents grounded knowledge of Azure SQL, plus there’s a new Azure SQL Dev Hub.

And the MSSQL extension for VS Code shipped a new landing page experience (preview), among other updates, so the first five minutes in the extension point you at the thing you actually came to do.

But that’s not all! If you’ve been following our drivers/SDK releases, you’ve hopefully noticed that we’ve kept up our incredible pace of driver updates. We released a new update to support Django 6.1 within 48 hours of the 6.1 release.  Arrow support in mssql-python now includes bulk copy. The big news this week is around connection string simplification, we added connection string keyword aliases to SqlClient, JDBC, ODBC, OLE DB, and go-mssqldb so the same keywords work across all of them. A connection string you write for one driver now works with all of our other drivers with fewer edits. And, perhaps the most anticipated, with the SQLAlchemy 2.1 release, mssyl-python is officially supported!

What next

Catch the sessions. The full SQLCon/FabCon Europe catalog, including the deep dives on everything above, is at aka.ms/sqlsessions.

Keep learning. New episodes of Data Exposed and other series land on the Azure SQL YouTube channel every week, and I’ll be dropping the video-version of this blog next week, if you want more of my personal commentary. Also, the SQL AI Developer certification covers the agentic workflows and modern tooling in this post, if you pass your certification, I want to know on LinkedIn (/amthomas46)

Tell us what to build next. The roadmap (now hosted at aka.ms/sqlroadmap) moves when you push on it. Bring your scenarios to the booth this week, or file them through the usual feedback channels.

–Anna