About Absolute DB

A database that fits on a postage stamp

~154 KB. Zero dependencies. Full SQL, vector search, graph database, and full-text search — all in a single C11 binary that starts in under two seconds.

What is Absolute DB

One binary. Every data primitive you need.

Absolute DB is a self-contained, embeddable database engine written in pure C11. The entire compiled binary weighs ~154 KB and carries zero runtime dependencies — no OpenSSL, no Boost, no libc extensions beyond the standard C library.

Despite its size, Absolute DB packs a full SQL engine (142 keywords, JOINs, CTEs, window functions, subqueries, prepared statements, stored procedures, and triggers), full ACID transactions with snapshot isolation and instant crash recovery, vector search via a built-in HNSW index (top-10 in 0.1 ms at 128 dimensions), a graph database engine (RELATE edges, BFS/DFS traversal, shortest paths at 1.3 billion edges/sec), and full-text search with BM25 ranking — all queried from the same SQL statement.

The server exposes a PostgreSQL v3 wire protocol on port 5433, so any existing Postgres client — psql, psycopg2, JDBC, pgAdmin, DBeaver, Prisma — connects without a single code change. A built-in JSON/HTTP REST API is included at no extra cost.

~154 KB binary Zero dependencies Full SQL — 142 keywords Vector search (HNSW) Graph database Full-text search (BM25) Full ACID Compliance PostgreSQL wire-compatible REST API + Web UI 2,164 tests passing Pure C11 Starts in <2 seconds
Why we built it

The problem with existing databases

Modern databases have become infrastructure projects. PostgreSQL is exceptional, but a fresh install pulls in hundreds of megabytes, requires a dedicated process, a DBA, and careful tuning before it handles production traffic. SQLite is tiny and brilliant for local storage, but it has no network server, no vector search, no graph traversal, and no real-time notification system.

The gap between "lightweight and embeddable" and "full-featured and production-ready" has always been enormous. Teams wanting vector similarity in their SaaS backend reach for a separate vector database. Graph queries require a separate graph engine. Full-text search demands Elasticsearch or Meilisearch. Every additional service multiplies operational burden, infrastructure cost, and failure surface area.

We built Absolute DB to close that gap completely. The design goal was a single binary that an embedded systems engineer can drop onto a 512 MB ARM board, that a SaaS developer can ship inside their Docker container without a second thought, and that an AI/ML team can use for both vector embeddings and structured metadata in one SELECT statement. No companion services. No configuration marathon. Type make all, run the binary, and you have a production-grade database.

Every number on this website — 2,164 tests, 304K+ inserts/sec, 0.1 ms vector search, 14.9M live notifications/sec — is measured live during make all. Run them yourself; they are part of the test suite.

Who uses it

Built for constrained environments and demanding workloads

Absolute DB was designed to excel where heavyweight databases are a liability and lightweight databases fall short.

🔌

Embedded Systems & IoT

Runs on ARM single-board computers, industrial PLCs, and resource-constrained edge devices. A ~154 KB binary with zero runtime dependencies fits into images where every megabyte counts. No separate database process to manage — link directly as a C library or run as a sidecar with negligible overhead.

☁️

SaaS Backends

Ship your entire data layer inside a single Docker container. Absolute DB's PostgreSQL wire compatibility means your existing ORM, query builder, and migration tooling all work without modification. Live queries push real-time updates to front-end clients at 14.9 million events per second — no separate pub/sub service required.

🤖

AI & ML Applications

Store and search vector embeddings from any model (up to 4,096 dimensions) alongside relational metadata in a single query. Build AI context graphs, retrieve nearest neighbours, filter by structured attributes, and rank with BM25 — all without stitching together three separate systems. Built-in HNSW delivers sub-millisecond top-k retrieval.

🌐

Edge Computing

Deploy Absolute DB at the network edge — in CDN PoPs, on-premise gateways, or regional micro-datacentres — to keep data processing close to users. The binary starts in under two seconds, uses minimal memory at idle, and requires no persistent configuration beyond the data file path. Built-in continuous replication keeps edge nodes in sync with the core.

🧪

Testing & CI/CD

Spin up a full-featured, PostgreSQL-compatible database in CI in under two seconds with no Docker image to pull and no external service to mock. Every pipeline gets an isolated, ephemeral database for integration and end-to-end tests. Tear it down by deleting a single file.

🏗️

Developer Tools & Desktop Apps

Bundle a complete relational + vector + graph database inside a desktop application, CLI tool, or developer utility without requiring users to install anything. Absolute DB links as a static C library or runs as a lightweight local server accessible over the PostgreSQL wire protocol — your pick.

🤝

Managed Service Providers

Replace expensive per-core SQL Server or Oracle licences with a flat-rate per-server model. Bundle Absolute DB into your managed hosting stack — backup, monitoring, and high availability are built in. PostgreSQL wire compatibility means zero changes to customer application code. Become a partner →

The company

Made in Australia

🇦🇺
Sydney, Australia

SupportCALL AU

Absolute DB is a product of SupportCALL AU, an Australian software company focused on building reliable, high-performance infrastructure tools. We believe software should do exactly what it says on the label — no hidden costs, no surprise dependencies, no licence nags on the free tier.

Absolute DB is developed and maintained by D.H.Maree, who designed the storage engine, query planner, vector index, graph layer, and wire protocol from scratch in C11. The project started as an internal tool for low-overhead data storage in constrained environments, and grew into a full-featured database after it kept outperforming far heavier alternatives in benchmark after benchmark.

The Community edition is and will remain free forever for individuals, open-source projects, and businesses under $1M annual revenue. Professional and Enterprise licences fund continued development, support, and the team that keeps the 2,164-test suite green.

Lead developer

DM

D.H.Maree

Systems programmer and database engineer based in Sydney, Australia. Designed and built Absolute DB from the ground up — a unified data platform delivering relational, vector, graph, and full-text capabilities in a single dependency-free binary.

davidmaree.com →
Contact

Get in touch

We respond to every email, usually within one business day (AEST, UTC+10/+11).

💬

General

Questions about Absolute DB, partnership enquiries, and everything else that doesn't fit another category.

💳

Sales

Professional licence keys, volume pricing, invoice-based payment, purchase orders, and trial extensions.

🏢

Enterprise

Custom SLAs, on-premise deployment consulting, dedicated support, MSA/DPA paperwork, and cluster licensing.

🛠️

Support

Bug reports, installation help, performance questions. Community edition users are welcome — we read every ticket.

🔒

Security

Responsible disclosure of vulnerabilities, CVE coordination, and security-related enquiries. PGP key available on request.

📄

Licensing

OEM embedding, redistribution rights, licence key recovery, and questions about what the Community licence permits.

🤝

Partnerships

MSP and reseller programme, channel partner tiers, co-marketing, cloud provider agreements, and OEM distribution.

For bug reports and feature requests, email support@absolutedb.com — we respond to every message.

Links

Find us around the web

Ready to try Absolute DB?

Download the Community edition free. No sign-up, no telemetry, no nags. Runs in 30 seconds.