Skip to content

A distributed database designed for correctness

Write applications with confidence using ACID transactions at any scale—simulation-tested, battle-proven, and open source.

Get started Learn more

FoundationDB Distributed Cluster Distributed • Transactional • Scalable

Everything you would expect

  • True ACID Transactions

    Multi-key, multi-range transactions with serializable isolation—the strongest guarantee possible. No eventual consistency surprises.

    Learn about transactions

  • Simulation Tested

    Our deterministic simulation framework tests millions of failure scenarios before every release. Bugs are found before they reach you.

    How it works

  • Horizontally Scalable

    Add nodes to increase capacity without application changes. Built for clusters from 3 nodes to hundreds of machines.

    Architecture

  • High Performance

    Millions of operations per second with single-digit millisecond latencies. Optimized for both read and write-heavy workloads.

    Getting started

  • Fault Tolerant

    Automatic failover and recovery. Your data survives hardware failures, network partitions, and datacenter outages.

    Operations guide

  • Multi-Model Foundation

    Build documents, graphs, tables, or queues on top of the ordered key-value store. The layer concept enables any data model.

    Data model

More than just a key-value store

Layered Architecture

Build any data model on top of FoundationDB's ordered key-value core. The Record Layer provides structured storage with indexing, while the Document Layer offers MongoDB-compatible APIs.

The layer concept lets you implement documents, graphs, tables, queues, or custom data structures—all with the same underlying ACID guarantees.

Explore the layers 

flowchart TB
    App[Your Application]
    App --> RL[Record Layer] & DL[Document Layer]
    RL & DL --> FDB[FoundationDB Key-Value Store]

Simulation Testing

Before any release, FoundationDB runs millions of simulated hours of operation, injecting every imaginable failure: disk corruption, network partitions, process crashes, and clock skew.

This deterministic testing framework catches bugs that would take years to appear in production—before they ever reach you.

Learn about simulation 

flowchart LR
    S((Simulation)) --> A[Network partitions]
    S --> B[Disk failures]
    S --> C[Process crashes]
    S --> D[Clock skew]
    S --> E[Byzantine faults]

Multi-Datacenter Replication

Deploy FoundationDB across multiple datacenters within up to two regions for disaster recovery and high availability. Satellite datacenters provide synchronous durability without WAN latency, while asynchronous replication between regions enables automatic failover.

Configure replication 

flowchart LR
    subgraph R1[Region 1]
        P1[Primary DC]
        S1[Satellite DC]
    end
    subgraph R2[Region 2]
        P2[Primary DC]
        S2[Satellite DC]
    end
    P1 <--> S1
    P2 <--> S2
    R1 -.->|async replication| R2

Trusted in production

Industry leaders rely on FoundationDB for mission-critical workloads. Over 10 years in production, handling millions of transactions per second.

See it in action

Swift as C++ Successor in FoundationDB — Konrad Malawski at Strange Loop 2023

Learn how FoundationDB is pioneering the use of Swift as a systems programming language, replacing C++ for new development while maintaining the database's legendary reliability.

Let's keep in touch

Made with and FoundationDB