Introduction
A programmable storage and stream engine for modern, distributed systems.
At its core, Litebase combines immutable, versioned storage with a transactional event stream, all built on top of object storage. This design enables it to scale horizontally across machines, services, and regions—without compromising on consistency or observability.
Unlike traditional databases that overwrite state or message queues that lose history, Litebase treats both data and change as first-class. Every write is recorded as an append-only event in the stream. The storage layer reflects this history as an ever-growing, versioned map of key-value records.
Built for Developers
With just a few simple REST APIs, developers can:
- Write records with full version history
- Read data at any point in time
- Watch changes as they happen in real time
There are no SDKs to install, no schema to configure. Litebase is designed to be stateless, language-agnostic, and easy to reason about.
Cloud-Native by Default
By separating compute and storage, Litebase leverages object storage as the source of truth, while using memory and SSD-based caching to accelerate active workloads. This means you can scale up ingestion, query, and replay—without worrying about managing disks or replicas.
Built for the Age of Agents and Dataflow
Whether you’re building AI pipelines, audit logs, stateful flows, or distributed agents, Litebase provides a foundation that’s:
- Immutable by design — every fact is durable
- Reproducible — stream events can always be replayed
- Composable — storage and stream are linked, not separated
To get started, head to the Quickstart guide. To learn how it all works, see Architecture.