Skip to content

DecentDB

Durable. Fast. Embedded.

DecentDB is an embedded, single-machine relational database engine focused on durable ACID writes and fast reads.

Features

  • ACID Transactions — Full durability with WAL-based persistence and snapshot isolation
  • Single Writer + Many Readers — Optimized for read-heavy workloads
  • PostgreSQL-Compatible SQL — JOINs (INNER, LEFT, RIGHT, FULL OUTER, CROSS, NATURAL), CTEs (including WITH RECURSIVE), window functions, aggregates (with DISTINCT), upsert, RETURNING, savepoints
  • Rich Data Types — INT64, FLOAT64, TEXT, BLOB, BOOL, DECIMAL, UUID, DATE, TIMESTAMP
  • Full-Text Substring Search — Trigram inverted index for LIKE '%pattern%' queries
  • Auto-Assigned Primary Keys — If a table has a single INT64 primary key column, omitting the value on INSERT will auto-assign an ID (INT/INTEGER/INT64/BIGINT are synonyms)
  • Foreign Keys — Referential integrity with CASCADE, SET NULL, RESTRICT
  • Generated Columns — GENERATED ALWAYS AS (expr) STORED for computed values
  • Temporary Objects — Session-scoped TEMP tables and views
  • JSON Support — Scalar functions, table-valued functions (json_each, json_tree)
  • Multiple Language Bindings — .NET, Go, Python, Node.js, JDBC
  • Cross-Platform — Linux x86_64/arm64 (including 64-bit Raspberry Pi OS), macOS, Windows

Releases

Releases are driven by Git tags and published via GitHub Actions:

  • Engine binaries (GitHub Releases): native Linux x86_64/arm64 (including 64-bit Raspberry Pi OS on Pi 3/4/5), macOS, and Windows builds. Releases
  • NuGet packages (.NET 10): DecentDB.AdoNet, DecentDB.MicroOrm, DecentDB.EntityFrameworkCore, DecentDB.EntityFrameworkCore.Design, DecentDB.EntityFrameworkCore.NodaTime

Quick Start

# Install DecentDB
nimble install decentdb

# Create a database
# Note: auto-increment works for a single INT64 PRIMARY KEY column (spelling INT/INTEGER/INT64 doesn’t matter).
decentdb exec --db=mydb.ddb --sql="CREATE TABLE users (id INTEGER PRIMARY KEY, name TEXT)"

# Insert data (id auto-assigned)
decentdb exec --db=mydb.ddb --sql="INSERT INTO users (name) VALUES ('Alice')"

# Query
decentdb exec --db=mydb.ddb --sql="SELECT * FROM users"

Use Cases

  • Offline-first desktop app — local relational cache for a UI-heavy app (fast reads, durable writes), with saveAs for backups/migration.
  • Music library / media server — trigram indexes for fast search across artist/album/track names and JSON metadata.
  • IoT / edge device data logger — append-only event table with native TIMESTAMP, periodic checkpoints, and snapshot exports.
  • Game tools / editors — temporary tables/views for import pipelines and fast iteration, with savepoints for “undo” style workflows.
  • Embedded analytics & reporting — GROUP BY/HAVING + window functions for dashboards on a single machine.
  • Config/state store for services — ACID transactions + foreign keys for consistent config + relational integrity.
  • ETL staging / ingestion — bulk-load CSV + generated columns for derived values and normalized search keys.
  • Search-heavy workloads — %pattern% queries accelerated by trigram indexes when you need substring matching.

Performance

  • Point lookups: P95 < 10ms
  • FK joins: P95 < 100ms
  • Substring search: P95 < 200ms
  • Bulk load: 100k records < 20 seconds

Getting Started