Download

Picodata

Open-source distributed database with shard-per-core architecture, PostgreSQL wire compatibility, and a Rust plugin SDK. Storage, ACID transactions, SQL, and a compute layer that runs inside the cluster.

Download Community Edition Documentation

What Picodata does

  • A linearised cluster-wide store for configuration, schema, and topology, built into the database.
  • SQL for single-node and cluster-wide queries. Two optimizers — distributed and local — push work close to the data.
  • Rust runtime libraries for networking, file I/O, green-thread scheduling, and embedded-storage access.
  • Cluster management via CLI, web UI, and Ansible roles.
  • Application lifecycle inside the cluster — versioning, dependencies, packaging, rolling deployment, zero-downtime upgrades.

Advantages

  • Durability you can defend

    ACID semantics, backed by a write-ahead log and synchronous replication. Topology and schema changes commit atomically through Raft consensus. No data loss when a single component fails.

  • Horizontal scale per core

    Shard-per-core architecture: each shard binds to a single CPU core with its own private memory, write-ahead log, and snapshot. No contention on shared resources. Add cores or nodes to grow capacity linearly, up to clusters of 10 000 cores.

  • Compute next to data

    Stored procedures and plugins run on the same nodes as the data they touch — no network round-trip for application logic. Code versions live cluster-wide; rolling upgrades happen with zero downtime.

  • Open source under BSD

    The Picodata core is BSD licensed and developed in the open. Read the source, file issues, send merge requests, fork it. The plugin SDK is open too — community plugins can be built outside the core team.

  • Plugin extensibility

    Rust plugins add new wire protocols, replication strategies, and compute primitives. They run against a shared SQL catalog. The Picodata team maintains four commercial plugins as proof of the model: Radix (Redis protocol), Sirin (Cassandra CQL), Ouroboros (cross-cluster replication), Franz (Kafka).

Supported systems

Picodata runs on x86-64 Linux:

  • RHEL 8 and 9 (and compatibles).
  • Debian 11, 12, 13.
  • Ubuntu 22.04, 24.04.
  • Fedora 42, 43.
  • Alpine (musl libc).
  • Docker images for any modern Linux host.

Open-source core, commercial plugins

The Picodata core (Community Edition) is BSD licensed and free to download, run, and modify. Source code lives at git.picodata.io/core/picodata.

Commercial plugins from the Picodata team are licensed per CPU core: Radix (Redis wire protocol), Sirin (Cassandra CQL), Ouroboros (cross-cluster async replication), Franz (Kafka reader). The SDK is open — the community can build, ship, and license its own.

Migration paths and pilot scope are on the Migration page.