El Bruno

Introduction

SQL Server 2025 is super cool 😎 — and naturally, I had to test out the brand-new vector capabilities! If you haven’t checked it out yet, start here:
🔗 https://learn.microsoft.com/en-us/sql/sql-server/what-s-new-in-sql-server-2025?view=sql-server-ver17

To explore these features, I leaned into one of my favorite playgrounds: eShopLite — a collection of sample scenarios for an eCommerce app with AI-powered Semantic Search. It supports multiple vector database options like Azure AI Search, Qdrant, ChromaDB, and even MCP Servers.

• 📦 Main repo here: https://aka.ms/eshoplite/repo
• 📦 SQL 2025 version here: https://aka.ms/eshoplite/sql2025

I wanted to create a new scenario using SQL 2025. But I hit a couple of snags:

1. The .NET Aspire NuGet package for SQL Server defaults to a SQL 2022 image 😬
2. Even if I spin up a SQL 2025 container — how do I enable Vector Search and configure Vector Indexes in the SQL DB Engine?

👉 Here’s how I solved it — in just 2 steps.

Step 1 – Create a SQL 2025 Server in .NET Aspire AppHost

Turns out, SQL 2025 is already available in the public image list in 🧊 Docker Hub !
The image I needed: mcr.microsoft.com/mssql/server:2025-latest. But Aspire uses 2022-latest by default…

🚀 Solution: Use the right image tag when creating the SQL server resource and that’s it!
( Note: David Fowler guided my in the right direction in this one! )
👉 See usage here.

var builder = DistributedApplication.CreateBuilder(args);

var sql = builder.AddSqlServer("sql")
    .WithLifetime(ContainerLifetime.Persistent)
    .WithImageTag("2025-latest")
    .WithEnvironment("ACCEPT_EULA", "Y");

var productsDb = sql
    .WithDataVolume()
    .AddDatabase("productsDb");

🎉 And just like that — I had a SQL Server 2025 instance running in a Docker container via Aspire!

Step 2 – Consume the SQL 2025 Server + Use Vector Search

Once my SQL 2025 instance was running, I needed to connect my app and enable vector search.

💡 The Aspire helper method .AddSqlServerDbContext<T>() doesn’t support vector search config (yet!)

So I turned to the NuGet package made for this EFCore.SqlServer.VectorSearch.

To make it work, I added this snippet to manually wire up the DB context. 👉 Full source here.

var productsDbConnectionString = builder.Configuration.GetConnectionString("productsDb");
builder.Services.AddDbContext<Context>(options =>
    options.UseSqlServer(productsDbConnectionString, o => o.UseVectorSearch()));

Boom 💥 — now my app can access vector indexes, and my Vector Entities can party 🎉.

🧠 Bonus: Semantic Search Like a Boss

With vector search enabled, I implemented Semantic Search like this. 👉 Full source here.

var embeddingSearch = embeddingClient.GenerateEmbedding(search, new() { Dimensions = dimensions });
var vectorSearch = embeddingSearch.Value.ToFloats().ToArray();

var products = await db.Product
    .OrderBy(p => EF.Functions.VectorDistance("cosine", p.Embedding, vectorSearch))
    .Take(3)
    .ToListAsync();

✨ So cool — it’s just a few lines of code, and now I’m doing AI-powered search directly inside SQL Server!

Conclusion

🚀 SQL Server 2025 + .NET Aspire = Vector-powered awesomeness 💪
I’ve got a vector database running in Docker, wired into my app, and now I’m semantically searching my products like it’s 2035 🤖

If you want to see this in action, I’ve got a 🔥 5-minute demo video coming soon on my YouTube channel.
👉 Subscribe here: https://youtube.com/elbruno

Stay tuned, stay curious — and go build cool stuff with .NET + AI! 💜🧠