datalevin/datalevin

A simple, fast and versatile Datalog database

What it solves

Datalevin is a durable Datalog database designed to be a simple, fast, and versatile alternative to SQL and graph databases. It addresses the complexity of SQL by providing a more declarative and composable query language, while avoiding the complex temporal features of some other Datalog implementations to maintain a familiar ACID-compliant behavior where deleted data is permanently gone.

How it works

Datalevin uses a cost-based query optimizer to achieve performance competitive with RDBMS like PostgreSQL and SQLite. It is built on a fork of LMDB (Lightning Memory-Mapped Database) to ensure high read performance and robust ACID transactions, supporting both Write-Ahead Logging (WAL) and asynchronous transactions for write-intensive workloads. It can operate as an embedded library (similar to SQLite) or in a networked client/server mode with Raft-based high availability and role-based access control (RBAC).

Who it’s for

It is designed for developers who prefer Datalog's declarative nature and implicit joins over SQL, as well as those needing a database that can handle relational, graph, document, and key-value data models. It is particularly useful for Clojure programs due to its native EDN data capabilities.

Highlights

  • Multi-model capabilities: Supports Datalog queries, document storage (JSON, EDN, Markdown) with automatic path indexing, vector database features via SIMD acceleration, and full-text search.
  • AI-native features: Includes a built-in local MCP server and supports in-DB embedding and text generation via integrated llama.cpp.
  • High Performance: Outperforms several alternatives in benchmarks for relational (JOB), graph (LDBC SNB), document (idoc), and logic (OpenRuleBench) workloads.
  • Broad Language Support: Provides libraries for Java, Python, Node.js, and Clojure, and is compatible with babashka pods.

Related

  • Project
    StarTrail-org/LEANN

    LEANN is an open‑source, ultra‑light vector database that stores almost no embeddings. It builds a graph of document chunks, recomputes embeddings on‑demand, and prunes the graph to use ~97 % less storage than traditional vector DBs while keeping search quality. It runs locally on a laptop, supports many data sources (files, emails, chat logs, live streams), offers HNSW/DiskANN back‑ends, integrates via MCP, and works with any OpenAI‑compatible LLM or embedding provider.

  • Project
    infiniflow/infinity

    An AI-native database providing high-speed hybrid search across dense and sparse embeddings, tensors, and full-text for LLM and RAG applications.

  • Project
    conductor-oss/conductor

    An open-source durable execution platform for microservices and AI agents that turns complex workflow graphs into persisted, inspectable, and recoverable executions.

  • Project
    lealone/Lealone

    A high-performance full-stack self-evolving agent that enables the creation of enterprise AI applications and full-stack apps via natural language and SQL-like service definitions.

  • Project
    milvus-io/milvus

    Milvus is a high-performance, distributed vector database built for scaling AI applications that require efficient search and organization of massive amounts of unstructured data.