Public Preview for SQL Server 2025 is Upon Us: DxEnterprise is Ready  

The future has arrived! Coinciding with the annual Microsoft Build conference taking place in Seattle, Microsoft announced the official public preview release of SQL Server 2025 earlier this summer. 

We tuned into Bob Ward and Muazma Zahid’s presentation at Microsoft Build, “SQL Server 2025: The Database Developer Reimagined,” to learn a bit more about the latest release of the trusted relational database management system (RDBMS). The presenting pair drove home the idea that SQL Server 2025 represents the most developer-focused update to the RDBMS in over a decade, but also made it clear that plenty of engineering resources have been poured into the SQL Server engine itself—citing upwards of 40 new features and enhancements.  

Three emphases of the release that our team at DH2i are especially excited about include the introduction of artificial intelligence (AI)-centric features, the commitment to platform and infrastructure flexibility, and of course, the architecture and enhancements to foster peak security, performance, and availability. 

SQL Server 2025 AI-Centric Features 

We’ve known since last year that Vector Search support was coming to SQL Server 2025, with AI being the most significant implication of the added features/support. Anyone can now download SQL Server 2025 Public Preview and try building SQL Server-backed GenAI applications themself, and they will be able to do so while leveraging the familiar T-SQL language through the process. 

Another concept that caught our eye during the presentation was what Ward keyed as “Operational” RAG. RAG refers to retrieval-augmented generation, which is the AI capability to enhance relevance and accuracy by supplementing information retrieval with generation. This allows large language models (LLMs) to access external data sources like knowledge bases, databases, and the open web to aid in the generation of the most informed and contextually relevant response possible. 

Operational RAG, as Ward described, improves on that, because with SQL Server 2025, “There is no separate vector store sitting outside of the SQL engine.” The vector store and data are co-located together so that operational databases can participate in the AI pattern. 

One more feature called out in the presentation was the useful capability of Structured Queries. This allows large language models to query structured data and take advantage of rich metadata and query optimization to deliver the best possible search results. 

SQL Server 2025 Platform and Infrastructure Flexibility 

Microsoft is certainly committed to allowing organizations to run SQL Server 2025 Public Preview and its associated infrastructure to support AI applications on-premises, in the cloud, or a hybrid combination of both. 

One notable phrase we clung on to from Ward and Zahid’s presentation was used to describe the flexibility of the new SQL Server AI tools, stating that they work from “ground to cloud to fabric.” So, whether you’re in Azure SQL or using SQL Server 2025 Public Preview, or the new SQL database in Fabric, these features exist anywhere you need them. 

SQL Server 2025 Security 

The introduction of AI in SQL Server 2025 certainly brings up some questions surrounding the holistic security of the platform—as GenAI apps require integration with 3rd party entities like LLMs. However, the Microsoft Build presentation tackled this concern head-on, providing some useful architecture diagrams to drive home the point that SQL Server 2025 Public Preview is as secure as ever.  

Specifically, the SQL Server engine is totally isolated from AI inferencing endpoints and LLMs. These components are connected using REST APIs. This architecture ensures maximum security and totally avoids the potential issue of AI processing models affecting operational workloads. For longtime SQL Server users, this architecture also means familiarity. Users wield total control over access for LLMs and AI inferencing endpoints using trusted and familiar SQL security. E.g. Roles, permissions, authentication, etc. 

SQL Server 2025 Performance and Availability 

Along with all the other enhancements and new features added, the Microsoft team touted a huge number of enhancements to performance and availability, and a dozen of those enhancements are focused solely on Availability Groups (AGs). AG enhancements and new features include: 

• Improved reliability for AG failovers
• AG tuning and diagnostics improvements
• Full database and differential backups on secondary replicas

Improved concurrency was another big performance enhancement to get the developer-heavy crowd excited at Microsoft Build. SQL Server 2025 Public Preview aims to ensure that when developers are connecting and running queries against SQL Server, that what they’re doing can remain concurrent and highly available without much, if any, involvement from their code. 

This is just a small taste of the many improvements and features added in the latest release. Take a look at Ward and Zahid’s full Microsoft Build presentation to learn more about the highlights of the public preview release of SQL Server 2025. 

Take AI Apps and Containers to Mission-Critical Environments with DH2i 

As a long-time partner and innovator in the SQL Server ecosystem, DH2i is excited to have achieved SQL Server 2025 Public Preview-Ready status with our DxEnterprise Smart High Availability software. 

DxEnterprise is a standalone replacement for high availability clustering solutions like Windows Server Failover Clustering and Red Hat Pacemaker. It offers a single unified high availability solution to manage all of your SQL Server across Windows, Linux, and Kubernetes. It also works in perfect tandem with SQL Server’s commitment to infrastructure flexibility by supporting high availability for on-premises deployments, cloud, or any mix. 

As a part of supporting this newest SQL Server release, DH2i integrates seamlessly with the AI-centric capabilities of SQL Server 2025 Public Preview. This includes maintaining high availability for databases that handle embeddings and function as vector stores, empowering organizations with the confidence to build highly available AI applications within their most mission-critical environments 

DxEnterprise will also continue to retain its full suite of container-centric capabilities with this preview release of SQL Server 2025—enabling organizations to: 

• Easily deploy SQL Server 2025 Availability Groups in Kubernetes using our SQL Server Operator, DxOperator
• Ensure the industry’s lowest downtime with fully automatic failover for SQL AGs in Kubernetes
• Span Windows, Linux, and Kubernetes with a single, unified AG

How to Get Started Creating Highly Available AI Apps

The public preview release of SQL Server 2025 marks an exciting new chapter in the evolution of Microsoft’s flagship RDBMS by opening the door to the Age of AI. DH2i’s DxEnterprise allows you to step through the door with confidence and flexibly deploy any mix of infrastructure and platforms that your organization needs to meet your unique SQL Server goals—all while prioritizing peak high availability, performance, and scalability.  

Anyone who is ready to get started building highly available AI applications with SQL Server 2025 Public Preview and DH2i can head to the following links to download the software to get started: 

DH2i DxEnterprise Free Trial: https://dh2i.com/trial/ 

SQL Server 2025 Public Preview: https://info.microsoft.com/ww-landing-sql-server-2025.html/  

If you’re not quite ready to try it yourself, get in contact with DH2i to schedule a free software demo. We’ll take a thorough look at the pain points you’re dealing with in your SQL Server environment, or your SQL Server 2025 modernization plan, and show you how DH2i can make life easier while improving overall efficiency and availability.  

If you’re interested in learning more about DH2i’s approach to smart high availability technology, get signed up for a one-on-one demo today.

Request a Demo

The author

Josh Achtemeier

Read more from this author