Features¶
Uni is an embedded, multi-model graph database that unifies graph, vector, document, and columnar analytics in a single engine. This section provides a user-focused view of what Uni does and when each feature is the right choice.
Summary¶
| Feature | What it gives you | Best for |
|---|---|---|
| Embedded & Serverless | Run Uni inside your process, no external server | Local apps, edge deployments, simple ops |
| Multi-Model Data Model | Graph + vector + document + columnar in one DB | Knowledge graphs, AI apps, mixed workloads |
| OpenCypher Querying | Familiar graph query language with extensions | Graph traversal and pattern matching |
| Columnar Analytics | Vectorized scans, filters, aggregates | Fast analytics without a separate warehouse |
| Window Functions | OVER clauses for ranking and running totals | Partitioned analytics in a single query |
| Vector Search | ANN search with HNSW/IVF_PQ/Flat | Semantic search, RAG, similarity |
| Full-Text + JSON Search | Text search over properties and JSON paths | Documents, metadata search |
| Schema & Indexing | Typed properties, constraints, and indexes | Performance, governance, stability |
| Graph Algorithms | Built-in centrality, clustering, pathing | Insights, scoring, routing |
| Transactions & Consistency | Snapshot isolation, single writer | Predictable reads, safe writes |
| Snapshots & Time Travel | Point-in-time snapshots + AS OF queries |
Auditing, reproducible analytics |
| Bulk Ingest | High-throughput loading with index rebuilds | Initial loads, large updates |
| Storage & Cloud Durability | Local + object-store storage with WAL | Low-ops durability on S3/GCS/Azure |
| APIs & Tooling | Rust + Python APIs, CLI, REPL | App integration and exploration |
When Uni Is the Right Choice¶
Use Uni when you want:
- A single embedded database for graph, vector, and document workloads.
- OpenCypher for graph traversal, combined with vector search or analytics.
- Local or object-store backed storage without managing a distributed system.
- Consistent reads with a simple single-writer model.
When Another Database Might Be Better¶
Consider alternatives when you need:
- Multi-writer distributed transactions across many machines.
- A fully managed, multi-tenant service with no embedded deployment.
- A pure analytics warehouse with SQL-first interfaces and federated query.
Why Uni Is a Strong Default for Graph + AI Workloads¶
Uni is designed to remove the usual trade-offs between graph traversal, vector search, and analytics. If your application needs those capabilities together, Uni tends to be the most direct and maintainable solution because it:
- Unifies graph, vector, document, and columnar queries in one engine.
- Runs embedded, avoiding a separate database service and network hops.
- Uses object-store friendly storage, so durability and scale are simple.
- Keeps the query surface small and consistent with OpenCypher.
If the constraints above do not apply, Uni is usually the most practical path to shipping graph + AI features in a single stack.