Core Concepts

Understand the fundamental concepts behind Uni's design and architecture.

### [Architecture](architecture.md) Layered design, component overview, and how the pieces fit together.

### [Data Model](data-model.md) Vertices, edges, properties, labels, and the property graph model.

### [Identity Model](identity.md) VID, EID, UniId, and how Uni identifies entities.

### [Indexing](indexing.md) Vector, scalar, hash, and full-text indexes for fast queries.

### [Concurrency](concurrency.md) Single-writer model, snapshot isolation, and concurrent readers.

Overview

Uni is built on three core principles:

1. Multi-Model Unification — Graph, vector, document, and columnar workloads in one engine
2. Embedded Simplicity — No separate server, network hops, or operational overhead
3. Cloud-Native Storage — Object store persistence with local caching for performance

Architecture at a Glance

┌─────────────────────────────────────────┐
│            Query Layer                   │
│   Parser → Planner → Executor           │
├─────────────────────────────────────────┤
│            Runtime Layer                 │
│   L0 Buffer | CSR Cache | Properties    │
├─────────────────────────────────────────┤
│            Storage Layer                 │
│   Lance Datasets | WAL | Indexes        │
├─────────────────────────────────────────┤
│            Object Store                  │
│      S3 | GCS | Azure | Local           │
└─────────────────────────────────────────┘

Next Steps

Start with Architecture for a comprehensive system overview.