Documentation
KiteDB documentation
Install KiteDB, define a schema, and query your graph from TypeScript, Python, or Rust. The internals section explains how the storage engine works, from the file format to MVCC.
Installation
Add the package for your language and open your first database file.
Quick start
Build a small social graph: define a schema, insert nodes, and traverse edges.
Guides
Schema definition
Define type-safe node and edge schemas
Queries & CRUD
Create, read, update, delete operations
Graph traversal
Navigate relationships in your graph
Vector search
Semantic similarity search with embeddings
Transactions
Atomic writes, snapshot isolation, conflicts, and bulk load
Performance checklist
Choose the fastest write path and config presets
Concurrency
Parallel readers, concurrent writers, and MVCC
API reference
Benchmarks
Internals
Architecture
How KiteDB is structured internally
Snapshot and delta
The core storage model
CSR format
How edges are stored for fast traversal
Single-file format
The .kitedb file layout
WAL and durability
Crash recovery and write-ahead logging
MVCC and transactions
Concurrent access and isolation
Key index
Fast node lookups by key
Performance
How KiteDB keeps reads and writes fast