DEV.co
Vector Databases · datalevin

datalevin

Datalevin is a durable Datalog database that combines SQL-like query performance with a more composable query language. It supports multiple deployment modes (embedded, client-server, or as a pod) and can function as a document store, vector database, or full-text search engine.

Source: GitHub — github.com/datalevin/datalevin
1.4k
GitHub stars
82
Forks
Clojure
Primary language
EPL-2.0
License (OSI-approved)

Key facts

Objective fields from the source. Values we can't verify are shown as “Unknown” rather than guessed.

FieldValue
Repositorydatalevin/datalevin
Ownerdatalevin
Primary languageClojure
LicenseEPL-2.0 — OSI-approved
Stars1.4k
Forks82
Open issues26
Latest release0.10.18 (2026-05-28)
Last updated2026-07-08
Sourcehttps://github.com/datalevin/datalevin

What datalevin is

Built on LMDB for high-read performance with ACID semantics and WAL support, Datalevin implements a cost-based query optimizer and Datalog query language compatible with Datomic. It offers native bindings for Java, Python, Node.js, and Clojure, with features including vector indexing, full-text search, embedded document indexing, and an MCP server for AI integrations.

Quickstart

Get the datalevin source

Clone the repository and explore it locally.

terminalbash
git clone https://github.com/datalevin/datalevin.gitcd datalevin# follow the project's README for install & configuration

Need it deployed, integrated, or customized instead? DEV.co ships production installs.

Best use cases

Embedded Database for Clojure/JVM Applications

Use as a drop-in durable state store for Clojure/Java applications requiring multi-dimensional querying without SQL boilerplate. Native EDN support and Datalog query language align well with Clojure semantics.

Document Database with Automatic Indexing

Store and query JSON/EDN documents (up to 2 GiB) with automatic path-based indexing. Useful for semi-structured data without needing explicit schema migrations or document-specific index management.

Vector Database for RAG/AI Workflows

Integrate embeddings, vector search, and full-text search in a single database with built-in llama.cpp support. MCP server enables direct AI client queries while keeping embedding generation and search co-located.

Implementation considerations

  • Embedded use requires understanding Datalog query semantics and schema design patterns; Datomic learning resources apply but Datalevin differences must be reviewed (cardinality definitions, schema structure, deletion semantics).
  • Client-server mode requires operational setup and maintenance of Raft cluster, RBAC, and WAL management; not a simple single-file database like SQLite when clustering is enabled.
  • Large documents (approaching 2 GiB) and vector indexing add memory and CPU overhead; benchmark transaction throughput and query latency against your specific workload before production use.
  • Java/Python/Node.js bindings exist but are maintained alongside the Clojure core; monitor binding stability and versioning alignment, especially for less common language bindings.
  • Vector search and full-text search features are built-in but performance against specialized engines (e.g., Elasticsearch, Vespa) is data-dependent; prototype with realistic datasets.

When to avoid it — and what to weigh

  • Heavy SQL Ecosystem Dependency — If your team and tooling are heavily invested in SQL, ORMs, and SQL-specific workflows, the learning curve for Datalog and schema differences from Datomic may incur friction.
  • Requires Temporal/Bitemporal Auditing — Datalevin explicitly rejects Datomic's temporal model—deleted data is gone. If you need built-in historical data tracking or point-in-time queries, use a different database.
  • Multi-Tenant SaaS at Massive Scale — While the README mentions Raft-based HA clustering, cluster design, operational complexity, and multi-tenant isolation strategies are not detailed. Large-scale multi-tenant deployments require careful review.
  • Existing Codebase with Strong Database Lock-In — Migrating from PostgreSQL, MongoDB, or other mature systems to Datalevin is non-trivial; lack of standardized migration tooling and smaller ecosystem may complicate adoption in established teams.

License & commercial use

EPL-2.0 (Eclipse Public License 2.0) is a copyleft open-source license that requires derivative works to be licensed under the same terms and makes source code available.

EPL-2.0 permits commercial use and distribution, including as embedded libraries in proprietary applications, provided modifications to Datalevin itself are published under EPL-2.0. Closed-source dependent applications are allowed. Review with legal counsel if significant modifications to Datalevin core are planned or if license compliance reporting is mandatory in your environment.

DEV.co evaluation signals

Editorial assessment — not user reviews. Directional, with an explicit confidence level.

SignalAssessment
MaintenanceActive
DocumentationAdequate
License clarityClear
Deployment complexityModerate
DEV.co fitGood
Assessment confidenceHigh
Security considerations

Server mode includes role-based access control (RBAC), but specific controls (authentication methods, encryption at rest, TLS in transit) are not described in the README. ACID semantics on LMDB fork provide transaction integrity. No security audit, CVE history, or penetration test results are cited. Evaluate threat model, compliance requirements, and security posture before handling sensitive data; request security documentation from maintainers.

Alternatives to consider

Datomic / Datomic Cloud

Mature, commercial Datalog database with extensive temporal features and cloud-native scaling, but proprietary, higher cost, and introduces the temporal semantics Datalevin deliberately avoids.

PostgreSQL (SQL + JSONB + Full-Text Search)

Established open-source RDBMS with SQL familiarity, mature ecosystem, and advanced features (JSONB, FTS, window functions). Lacks Datalog composability and recursive rules, but proven at scale.

MongoDB + Vector Search (Atlas)

Document-oriented, mature ecosystem, built-in vector search (Atlas Search). No Datalog; requires schema-on-read management and operational complexity for sharding; higher operational overhead than embedded solutions.

Software development agency

Build on datalevin with DEV.co software developers

Download the source, run the examples, and prototype with your data. For clustering, HA, and production security posture, engage the maintainers or conduct a vendor security review. Consider your team's Datalog familiarity and migration effort before committing.

Talk to DEV.co

Related open-source tools

Surfaced by semantic similarity across the DEV.co open-source index.

Related on DEV.co

Explore the category and the services that help you build with it.

datalevin FAQ

Can I use Datalevin as a drop-in replacement for SQLite?
Partially. Both are embeddable and durable, but Datalevin uses Datalog queries (not SQL), different schema patterns, and is Clojure-first. Migrating existing SQL schemas and queries requires rewrite. Consider Datalevin if you value Datalog composability; use SQLite if SQL compatibility is paramount.
What is the storage overhead and disk size limit?
Built on LMDB, which is memory-mapped; storage is efficient and scales to very large datasets. Individual documents are limited to <2 GiB. Total database size is not explicitly capped in the README; review LMDB documentation and test with your data size expectations.
How does clustering and high availability work?
Datalevin supports Raft-based HA clustering with read replicas and RBAC. The README does not detail failover time, consistency guarantees, or operational procedures; consult the server documentation or vendor for production HA design patterns.
Is Datalevin suitable for mobile or edge deployment?
Embedded mode (library) is lightweight and could work on resource-constrained environments, but no benchmarks, binary size, memory footprint, or mobile-specific bindings are mentioned. Prototype on target hardware before committing to production mobile use.

Software development & web development with DEV.co

From first prototype to production, DEV.co delivers software development services around tools like datalevin. Our software development agency staffs experienced software developers and web developers for custom software development, web development, integrations, and ongoing support across vector databases and beyond.

Ready to Evaluate Datalevin?

Download the source, run the examples, and prototype with your data. For clustering, HA, and production security posture, engage the maintainers or conduct a vendor security review. Consider your team's Datalog familiarity and migration effort before committing.