A production-ready data access platform for .NET applications.
RepoDB is a high-performance data productivity platform for .NET developers. At its core is the popular hybrid-ORM library for .NET with clean and easy APIs.
It provides the flexibility to work the way you want — all through the IDbConnection interface. Write raw SQL when you need absolute control, or use the fluent APIs for more productivity, and switch seamlessly between both without sacrificing performance or maintainability.
RepoDB solves a complex problem in the .NET data access space: making bulk operations simple. It is designed to efficiently move millions of records across different database providers. Imagine migrating massive datasets from legacy databases to modern, cloud-native platforms — efficiently, reliably, and with high performance.
It goes beyond the ORM with enterprise-grade capabilities for building, operating, and scaling with confidence:
- 📦 Bulk Operations — High-performance bulk inserts, updates, merges, and deletes built for demanding production workloads.
- 🔄 Data Replication — Scalable data movement and synchronization across multiple database platforms.
- 📊 Telemetry — Immediate visibility into execution times, failures, and application behavior, with minimal configuration.
It stays close to the metal and remaining lightweight:
- 👌 Easy to Use — All operations are extension methods on IDbConnection. Open a connection and you're ready to go.
- 🚀 High Performance — Compiled expressions are cached and reused. RepoDB understands your schema to generate the most efficient execution path ahead of time.
- 🧠 Memory Efficient — Object properties, execution contexts, mappings, and SQL statements are extracted once and reused throughout the lifetime of your application.
- 🔀 Hybrid — Use fluent methods for everyday CRUD, drop down to raw SQL for complex queries, or mix both — all within the same connection.
- 🏆 Battle-Tested — Backed by thousands of unit and integration tests, and used in production systems worldwide.
- 🆓 Always Free — Apache 2.0 licensed, forever open source.
It bridges different database systems with a single, unified connectivity layer:
- 🔌 Universal Connectivity — Works with any provider that implements IDbConnection, from SQL Server and PostgreSQL to ClickHouse, DB2, Oracle, and more.
- 🌉 Cross-Provider Data Movement — Move data between different database engines without writing hand-rolled conversion logic.
- 🧩 Consistent API Surface — The same fluent methods and conventions work identically regardless of the underlying provider.
- ⚙️ Provider-Native Optimizations — Each connector is tuned to exploit the bulk and native capabilities of its target database.
RepoDB is evolving from an ORM into a broader data productivity platform, with Data Engineering capabilities — moving data between providers, on-premise or cloud-native — on the roadmap.
Read more about this through this article.
Fluent operations (Query, Insert, Merge, Delete, Update, and more), batches and bulks are supported for DB providers mentioned below. Choose your database and follow the quick-start guide:
- ClickHouse
- DB2
- EnterpriseDB
- Firebird
- MariaDB
- MySQL
- Oracle
- PostgreSQL
- SAP HANA
- SQL Server
- SQLite
- Vertica
While raw SQL execution methods work with any ADO.NET-compatible provider:
Explore individual features in the documentation.
We have envisioned the following databases to be supported to further grow the ecosystem.
RepoDB ships as a core package plus provider-specific packages for each supported database, with optional bulk-operations add-ons.
See the full package list and build status.
Every supported database has a dedicated bulk-operations add-on (BulkInsert, BulkMerge, BulkUpdate, BulkDelete), each built on top of that provider's own fastest data-loading mechanism, with sync and async variants:
| DB Provider | Sync | Async | Class |
|---|---|---|---|
| ClickHouse | ✅ | ✅ | ClickHouseBulkCopy / OPTIMIZE_ON_INSERT |
| DB2 | ✅ | ✅ | DB2BulkCopy Native (Sync) / Db2BulkArrayBinder (Async Native) |
| EnterpriseDB | ✅ | ✅ | EDBBulkCopy (via RepoDb.Connector.EnterpriseDb) via Npgsql Native |
| Firebird | ✅ | ✅ | FirebirdCommandBatcher (via FbBatchCommand) Native |
| MariaDB | ✅ | ✅ | MariaDbBulkCopy (via RepoDb.Connector.MariaDb) Native |
| MariaDB Connector | ✅ | ✅ | MariaDbBulkCopy (via RepoDb.Connector.MariaDbConnector) Native |
| MySQL | ✅ | ✅ | MySqlBulkCopy (via MySqlBulkLoader Native) |
| MySQL Connector | ✅ | ✅ | MySqlConnectorBulkCopy Native |
| Oracle | ✅ | ✅ | OracleBulkCopy Native / OracleBulkArrayBinder (Async Native) |
| PostgreSQL | ✅ | ✅ | NpgsqlBinaryImporter Native |
| SAP HANA | ✅ | HanaBulkCopy Native (Sync) / AsyncOverAsync or SapHanaCommandBatcher |
|
| SQL Server | ✅ | ✅ | SqlBulkCopy Native |
| Vertica | ✅ | VerticaBulkCopy (via VerticaCopyStream) / Async over Sync |
RepoDB uses ADO.NET's native coercion by default, keeping type mismatches visible and explicit. To enable automatic conversion:
GlobalConfiguration
.Setup(new Options.GlobalConfigurationOptions()
{
ConversionType = ConversionType.Automatic
})
.UseSqlServer();RepoDB sits between a micro-ORM and a full ORM. Each tool below makes different tradeoffs — pick the one that fits your project:
| RepoDB | Dapper | Entity Framework | |
|---|---|---|---|
| Abstraction level | ✅ Hybrid — fluent CRUD + raw SQL | ❌ Micro-ORM — raw SQL mapping | ✅ Full ORM — LINQ, change tracking |
| Fluent CRUD API | ✅ Yes (Insert, Query, Update, Delete, Merge, more) | ❌ No — SQL per call | ✅ Yes, via LINQ/DbSet |
| Raw SQL | ✅ Yes, mixed freely with fluent calls | ✅ Yes — its core model | ✅ Yes, via FromSql |
| Change tracking | ❌ None | ❌ None | ✅ Yes |
| Migrations | ❌ None built-in | ❌ None built-in | ✅ Yes (EF Migrations) |
| Native Bulk | ✅ Built-in, cross-provider | ❌ Via extensions | ❌ Via extensions/third-party |
| Insights / telemetry | ✅ Built-in (RepoDb.Telemetry.Default) | ❌ None built-in — manual or third-party (e.g. MiniProfiler) | ✅ Built-in logging/interceptors; OTel via community packages |
| Performance | ✅ Close to raw ADO.NET | ✅ Close to raw ADO.NET | ❌ Overhead from tracking/materialization |
| Best fit | ✅ EF-like productivity without losing SQL control | ✅ Thinnest possible SQL-to-object mapper | ✅ Rich object graphs, LINQ, migrations |
Dapper and Entity Framework are both excellent, mature tools — this reflects design tradeoffs, not a ranking.
RepoDB includes opt-in, drop-in telemetry via RepoDb.Telemetry.Default. Enable it once at startup and every operation (Insert, Query, Update, Delete, etc.) is captured and published to your insights collector automatically — no custom ITrace required.
It comes with great and simple dashboards visualization.
Simply docker compose up -d the docker-compose.yml and .env files and integrate your code.
GlobalConfiguration
.Setup(new GlobalConfigurationOptions { UseRegisteredGlobalTraces = true })
.UseDefaultTelemetry(new DefaultTelemetryOption("<YOUR_APPLICATION_NAME>")
{
Host = "https://your-collector-host",
ApiKey = "YOUR_API_KEY",
Group = "<YOUR_APPLICATION_GROUP>",
Frequency = TimeSpan.FromSeconds(1)
});It's intentionally lightweight rather than OTel-based, keeping RepoDB's thin, fast footprint intact. See the package README for configuration options, the full OTel rationale, and the roadmap.
If your organization is evaluating or adopting RepoDB for production use, keep the following in mind:
- 🖥️ Run the benchmark on your environment — Published numbers (e.g., the benchmark suite) are produced on infrastructure local to the project (local Docker containers, CI runners, contributor machines). Hardware, network topology, database configuration, cloud provider, virtualization/container overhead, and data volume all materially affect performance, so clone the repository and run the relevant benchmark project(s) against infrastructure and data shapes that mirror your own workload before relying on any published figure.
⚠️ Be aware of the limitations — RepoDB is a micro-ORM built for advanced use cases and has documented, provider-specific limitations (e.g., composite keys, JOINs, cache invalidation, and behaviors specific to SQL Server, Oracle, DB2, and other providers). Review these against your intended usage before committing to production.- 🔒 Security is best-effort, not SLA-backed — There is no dedicated security team or response SLA. Read our Security Policy before adopting RepoDB in a security-sensitive context — it covers how to report a vulnerability, which versions get fixes, and the RepoDB-specific risk areas (raw SQL execution, reflection-based provider internals, unscanned third-party driver dependencies) along with how each is mitigated.
- 🤝 Understand our contributions and support policy — RepoDB is primarily maintained by a single individual, not an organization. Community contributions are welcome and encouraged (see Contributing), and all support is free (for now), but it follows a documented support policy with defined hours and channels rather than an enterprise SLA. Plan your adoption and escalation paths accordingly, and consider contributing back or arranging consultancy for dedicated needs.
- 🚌 Factor in the bus factor — RepoDB has no formal governance structure today; entitlement, succession, and legal/trademark ownership are not yet defined (see Governance). Account for this single-maintainer risk in your continuity and vendor-risk planning.
- 🔢 No formal versioning/breaking-change policy — RepoDB does not publish a documented SemVer or deprecation policy, and (per the Security Policy) only the latest release of each package receives fixes — there is no backport process to older majors/minors. Pin exact package versions in production and review release notes before upgrading, rather than assuming strict backward compatibility across releases.
We welcome contributions of all kinds — code, docs, bug reports, and ideas.
- Browse for-grabs issues and submit a PR.
- File a new issue to start a discussion.
- Contribute to the documentation site.
- Blog about it, share it, or simply give us a ⭐
- GitHub Issues — bug reports and feature requests.
- Microsoft Teams — live Q&A and community chat.
- GitHub Discussions — ask questions and share ideas.
- X / Twitter — news and updates.
Read our contributing page for more.
- Building the Solutions
- Coding Standards
- Issuing a Pull Request
- Reporting an Issue
- Support Policy
- Limitations
Thanks to all contributors and to Scott Hanselman for featuring RepoDB.
Tools and projects that make RepoDB possible: GitHub, GitHub Actions, Docker, Microsoft Teams, Moq, MSTest, NuGet, RawDataAccessBencher, Shields, Microsoft.Data.SqlClient, Microsoft.Data.Sqlite, System.Data.SQLite.Core, MySql.Data, MySqlConnector, Npgsql, FirebirdSql.Data.FirebirdClient, Net.IBM.Data.Db2, Oracle.ManagedDataAccess.Core, ClickHouse.Driver, RepoDb.Connector.MariaDb, RepoDb.Connector.MariaDbConnector, Vertica.Data, RepoDb.Connector.EnterpriseDb, Sap.Data.Hana.Net.v6.0.
Apache-2.0 — Copyright © 2018 Michael Camara Pendon