DuckDB Entity Framework Core Integration

This provides a seamless integration between Entity Framework Core and DuckDB, allowing you to use DuckDB as your database while maintaining the familiar EF Core API.

⚠️ Native library dependencies not included This package does not include the DuckDB or SQLite native libraries. You must add the appropriate driver packages to your project separately — see the Installation section below.

Features

• Simple API: Just use UseDuckDb() similar to UseSqlite() or UseSqlServer()
• Automatic SQL Translation: Converts EF Core generated SQL to DuckDB-compatible SQL
• Full EF Core Support: Works with LINQ queries, navigation properties, and all EF Core features
• Performance: Leverages DuckDB's columnar storage for fast analytical queries on Parquet files with significant performance improvements over SQLite
• Developer-Friendly API: The extension replaces fragile, manual setups with a clean, fluent API. This reduces code complexity, improves maintainability, and accelerates development.

Performance

The performance improvements are significant when using DuckDB over SQLite for analytical queries:

Database	OData Query Performance
SQLite	7.61s - 8.83s
DuckDB	800ms - 1s

Performance Gain: ~8-10x faster 🚀

DuckDB's columnar storage and vectorized execution engine provide substantial performance benefits for analytical workloads, making it an excellent choice for data analysis and reporting scenarios.

Installation

Add the main package to your project:

<PackageReference Include="EnergyExemplar.EntityFrameworkCore.DuckDb" Version="*" />

Native Binary Dependencies

The native library requirements differ by target framework:

.NET 8 (net8.0)

The net8.0 build of this package includes the native DuckDB and SQLite binaries transitively via DuckDB.NET.Data.Full and SourceGear.sqlite3. No additional packages are required.

.NET 10 (net10.0)

The net10.0 build does not include native binaries. You must supply them explicitly.

DuckDB native binaries — add one of:

<PackageReference Include="DuckDB.NET.Data.Full" Version="1.5.2" />

SQLite native binaries — add one of:

• Microsoft.EntityFrameworkCore.Sqlite (the full package, not .Core) — if you are already referencing it in your project, its transitive dependency on SQLitePCLRaw.bundle_e_sqlite3 covers SQLite binaries automatically.
• Or add a standalone SQLite bundle explicitly:

<PackageReference Include="SourceGear.sqlite3" Version="3.50.4.5" />

Why the difference? The net8.0 build bundles binaries to avoid breaking existing consumers. The net10.0 build intentionally omits them so you can choose the binary packages appropriate for your platform and deployment environment.

Versioning

Starting with version 10.0.0, this package adopts a versioning scheme aligned with the .NET and EF Core major version it targets:

Package version	Target framework	EF Core version
10.0.x	net10.0	10.x
8.0.x	net8.0	8.x
1.x (legacy)	net8.0	8.x

When referencing the package, pin to the major version that matches your target framework:

<!-- .NET 10 projects -->
<PackageReference Include="EnergyExemplar.EntityFrameworkCore.DuckDb" Version="10.0.*" />

<!-- .NET 8 projects -->
<PackageReference Include="EnergyExemplar.EntityFrameworkCore.DuckDb" Version="8.0.*" />

Upgrading from 1.x? If you are currently on a 1.x release (net8), move to the 8.0.x range. No API changes are required — only the version pin needs updating.

Basic Usage

1. Lambda-Based Configuration (Recommended)

using EnergyExemplar.EntityFrameworkCore.DuckDb;
using Microsoft.EntityFrameworkCore;

var options = new DbContextOptionsBuilder<MyDbContext>()
    .UseDuckDb(opts =>
    {
        opts.ConnectionString = "DataSource=C:/data/mydb.duckdb";
        opts.MemoryLimitGB = 4; // Optional: set DuckDB memory limit to 4GB
        opts.FileSearchPath = "C:/data/"; // Optional: set file_search_path for relative views
    })
    .Options;

using var context = new MyDbContext(options);
var data = await context.MyEntities.Where(e => e.Name.StartsWith("Gen")).ToListAsync();

2. Using Dependency Injection (ASP.NET Core)

// In your Program.cs or Startup.cs
services.AddDbContext<MyDbContext>(options =>
    options.UseDuckDb(opts =>
    {
        opts.ConnectionString = builder.Configuration["DuckDb:ConnectionString"];
        opts.MemoryLimitGB = 8;
        opts.FileSearchPath = builder.Configuration["DuckDb:FileSearchPath"];
    })
);

3. Using a Configuration Object

var duckDbOptions = new DuckDbConnectionOptions
{
    ConnectionString = "DataSource=C:/data/mydb.duckdb",
    MemoryLimitGB = 16,
    FileSearchPath = "C:/data/"
};

var options = new DbContextOptionsBuilder<MyDbContext>()
    .UseDuckDb(duckDbOptions)
    .Options;

4. Parquet File Only (No .duckdb Database)

var options = new DbContextOptionsBuilder<MyDbContext>()
    .UseDuckDbOnParquet("C:/data/mydata.parquet")
    .Options;

Tip: Change tracking is disabled by default for maximum performance. To enable it, pass noTracking: false to UseDuckDbOnParquet() or set NoTracking = false on DuckDbConnectionOptions.

Migration from Previous Implementation

Before (Manual Setup)

Manual DbConnection management, custom interceptors, and scattered configuration.

After (Using UseDuckDb)

var options = new DbContextOptionsBuilder<MyDbContext>()
    .UseDuckDb(opts =>
    {
        opts.ConnectionString = "DataSource=C:/data/mydb.duckdb";
    })
    .Options;

Configuration Options

DuckDbConnectionOptions Properties

• ConnectionString (required): Path to the DuckDB database file (e.g., DataSource=mydb.duckdb)
• FileSearchPath (optional): Base directory for resolving relative paths in DuckDB views
• MemoryLimitGB (optional): Memory limit for DuckDB operations (in GB)
• NoTracking (optional): If true (the default), all queries will be executed with AsNoTracking(). Set to false to enable change tracking.

Example: Complete Application

using Microsoft.EntityFrameworkCore;
using Microsoft.Extensions.DependencyInjection;
using Microsoft.Extensions.Hosting;
using EnergyExemplar.EntityFrameworkCore.DuckDb;

var host = Host.CreateDefaultBuilder(args)
    .ConfigureServices((context, services) =>
    {
        services.AddDbContext<MyDbContext>(options =>
            options.UseDuckDb(opts =>
            {
                opts.ConnectionString = context.Configuration["DuckDb:ConnectionString"];
                opts.MemoryLimitGB = 8;
                opts.FileSearchPath = context.Configuration["DuckDb:FileSearchPath"];
            })
        );
        services.AddScoped<IDataService, DataService>();
    })
    .Build();

await host.RunAsync();

public class DataService : IDataService
{
    private readonly MyDbContext _context;
    public DataService(MyDbContext context) => _context = context;
    public async Task<List<MyEntity>> GetDataAsync()
        => await _context.MyEntities.Where(e => e.IsActive).ToListAsync();
}

Advanced Scenarios

Multiple Contexts

services.AddDbContext<InputContext>(options =>
    options.UseDuckDb(opts =>
    {
        opts.ConnectionString = "DataSource=input.duckdb";
    })
);
services.AddDbContext<OutputContext>(options =>
    options.UseDuckDb(opts =>
    {
        opts.ConnectionString = "DataSource=output.duckdb";
    })
);

Parquet-Only Context

services.AddDbContext<MyParquetContext>(options =>
    options.UseDuckDbOnParquet("C:/data/mydata.parquet")
);

Troubleshooting

1. Memory Issues: Set MemoryLimitGB in the options lambda or object.
2. Path Issues: Ensure FileSearchPath is set correctly for relative paths in views.
3. Write Operations: The integration is read-only; any write operation will throw NotSupportedException.
4. Performance: DuckDB excels at analytical queries but may be slower for OLTP workloads.

Benefits Over Manual Configuration

1. Cleaner Code: No need to manually configure interceptors and options
2. Type Safety: Configuration options are strongly typed
3. Consistency: Ensures all required components are properly configured
4. Maintainability: Changes to DuckDB integration are centralized
5. Testability: Easy to mock or substitute in tests

Notes

• The implementation uses SQLite as the base provider but intercepts and translates all SQL to DuckDB
• Query tracking is disabled by default for better performance (can be overridden via NoTracking = false)
• All DuckDB dependencies are automatically registered when using the extensions

Custom Conventions (Optional)

The library includes optional EF Core conventions that can simplify your model configuration by automatically mapping entity and property names directly to table and column names.

Available Conventions

1. EntityNameAsTableNameConvention: Automatically sets table names to match entity class names
2. PropertyNameAsColumnNameConvention: Automatically sets column names to match property names
3. CustomConventionSetBuilder: Applies both conventions together

Usage

using EnergyExemplar.EntityFrameworkCore.DuckDb.Conventions;
using Microsoft.EntityFrameworkCore;
using Microsoft.EntityFrameworkCore.Metadata.Conventions.Infrastructure;

// Option 1: Apply conventions when configuring DbContext
var options = new DbContextOptionsBuilder<MyDbContext>()
    .UseDuckDb(opts =>
    {
        opts.ConnectionString = "DataSource=mydb.duckdb";
    })
    .ReplaceService<IProviderConventionSetBuilder, CustomConventionSetBuilder>()
    .Options;

// Option 2: Apply in dependency injection
services.AddDbContext<MyDbContext>(options =>
{
    options.UseDuckDb(opts =>
    {
        opts.ConnectionString = configuration["DuckDb:ConnectionString"];
    })
    .ReplaceService<IProviderConventionSetBuilder, CustomConventionSetBuilder>();
});

Benefits

• Reduced Boilerplate: No need to manually configure table and column names
• Consistency: Ensures naming conventions are applied uniformly across your model
• DuckDB Compatibility: Particularly useful when working with DuckDB views or Parquet files where column names must match exactly
• Simplified Migrations: When entity/property names match database schema, there's less configuration needed

Example Scenario

Without conventions:

public class CustomerOrder
{
    public int OrderId { get; set; }
    public string CustomerName { get; set; }
}

// Requires manual configuration
modelBuilder.Entity<CustomerOrder>()
    .ToTable("CustomerOrder")
    .Property(e => e.CustomerName)
    .HasColumnName("CustomerName");

With conventions (automatically applied):

public class CustomerOrder
{
    public int OrderId { get; set; }
    public string CustomerName { get; set; }
}
// No additional configuration needed!

When to Use

These conventions are particularly helpful when:

• Your entity and property names already match your database schema
• You're working with existing Parquet files or DuckDB views
• You want to reduce EF Core configuration code
• You prefer convention over configuration approach

Note: These conventions are optional. If you need different table/column names than your entity/property names, simply don't apply the conventions and use standard EF Core configuration.

License

This project is licensed under the MIT License - see the LICENSE file for details.

Third-Party Licenses

This project depends on the following third-party libraries:

• Microsoft.EntityFrameworkCore - Licensed under the MIT License. See Entity Framework Core repository for details.
• Microsoft.EntityFrameworkCore.Sqlite - Licensed under the MIT License. See Entity Framework Core repository for details.
• DuckDB.NET.Data.Full - Licensed under the MIT License. See DuckDB.NET repository for details.