10 February 2025

What lies ahead: AI integration and performance boosts in SQL Server 2025

In this blog, Lee Brownhill, Principal DBA at Cloud Rede, shares his thoughts on SQL Server 2025, AI integration, and performance improvements. As someone who works closely with clients, Lee brings a grounded perspective on how these new features could impact day-to-day SQL operations. 

SQL Server 2025 is on the horizon and was showcased by Bob Ward and his team in November 2024 at Microsoft Ignite. The headline features are heavily slanted toward all things AI, no surprises there!

Key features of SQL Server 2025

  • AI Built-In: Calls directly from SQL Server to AI services, vector searches, and dedicated vector datatypes.
  • Fabric Integration: Fabric mirroring will be available in SQL 2025, further protecting those operational workloads from analytical queries.
  • Database Development Improvements: GraphQL, JSON expanded further with JSON data types and indexes.
  • Security Improvements: Managed identities for both incoming and outbound connections.
  • Database Engine Enhancements: Optimised locking, batch mode improvements, and further enhancements for parameter-sensitive plan optimisation.

These features paint a picture of SQL Server 2025 as a version heavily influenced by AI technologies, but how much of that will truly impact the day-to-day operations for most users?

Initial thoughts on SQL Server 2025

As someone on the ground, working with our clients, I can’t help but think to myself who this version is for. From my experience, our clients’ two biggest wishes for SQL Server, for most projects and workloads, are stability and performance. The AI elements of the update are ‘cool’ and ‘exciting,’ but they’re still outside the scope of most workloads I encounter—at least for now.

With that said, my eye naturally goes straight to the Database Engine Enhancements…

Personally and professionally, I’m very interested in the optimised locking, which has now been in Azure SQL DB for a while. “Forget about lock escalation,” and any feature that claims to improve overall concurrency within the database is always going to get my attention. Let’s hope it’s a feature worth talking about once it’s released!

Batch mode improvements also sound exciting. When helping clients overcome performance issues, batch mode has really helped us get over the line at times. The technology can be truly impressive for the correct application and still amazes me when I see it take a query and totally revamp its execution method and time.

Further enhancements have been promised for parameter-sensitive plan optimisation. Now, this feature is something I want Microsoft to continue to get behind. As most of us working with SQL Server know, parameter sniffing issues can be commonplace, and I appreciate how tricky this must be for Microsoft to ‘solve’ completely. Fingers crossed the enhancements are as good as we all hope, and those parameter sniffing issues will be a thing of the past.

Will SQL Server 2025 be a game-changer?

Is SQL Server 2025 the version we’ll be talking about for years to come? I’m not so sure yet, and only time will tell. A lot of the features highlighted may benefit niche customers performing niche tasks. It would have been nice to see more Database Engine Enhancements mentioned, but who knows? There could be more in this area in the future—fingers crossed.

Internally, here at Cloud Rede, we’ll be testing out these touted features with both interest and hope, as soon as possible. Any updates to SQL Server are a good thing, and any additional tools we can utilise for our clients’ needs are always a positive in my book.

But will these new features help the average workload using SQL? In my experience, people aren’t using AI as much as the press would have us believe—not yet, at least, and certainly not anywhere near their database.

If my interactions with clients over the past few years are anything to go by, people want stability and performance features. That’s the number one priority, no questions asked. We still face issues like query regression and spills to disk that suddenly start happening—so, while AI might have its place, I fear it’ll be sitting in the cheap seats, wondering if there’s anything it can help with. Anyone know if AI can make a coffee yet?

Want to stay ahead of the curve with your SQL Server strategy?

Let’s discuss how Cloud Rede can help you optimise performance, stability, and performance tuning. Get in touch with us today, and let’s ensure your business is ready for whatever the future of SQL has in store!

Tags:

AI | Fabric Integration | GraphQL | JSON | Microsoft Ignite | SQL Server 2025

Share:

Let’s Talk

Have a SQL challenge? We can help.

"*" indicates required fields

This field is for validation purposes and should be left unchanged.