Documentation

Frequently Asked Questions

Answers to the most common questions about Absolute DB — what it is, how it compares to other databases, licensing, compatibility, and how to get started.

General

What is Absolute DB?

Absolute DB is a modern, AI-native database written in pure C11 with zero external library dependencies. It delivers SQL:2023 compliance, native vector search, full-text BM25, graph traversal, time-series hypertables, and geospatial indexing in a single binary. The full server binary is approximately 20 MB — comparable in size to PostgreSQL but with substantially more built-in features. The embedded lite binary is approximately 2 MB.

Absolute DB is designed to be the single source of truth for modern applications: replacing separate databases for OLTP, vector search, full-text search, caching, time-series, and messaging with one unified engine.

How does it compare to PostgreSQL?

Absolute DB speaks the PostgreSQL wire protocol v3 on port 5433 (also 5432 with -p 5432), so existing PostgreSQL drivers, ORMs, and tools connect without modification. SQL:2023 conformance is 100% (150/150 tests pass).

Key differences: Absolute DB ships vector search (HNSW), BM25 full-text, graph traversal, time-series hypertables, gRPC, native TLS 1.3, post-quantum cryptography, and built-in message queues in the core binary — capabilities that require separate extensions or entirely separate systems in PostgreSQL. Absolute DB also has no configuration files required to get started — a single binary runs immediately.

How does it compare to SQLite?

The absdb-lite embedded binary (~154 KB) targets the same use cases as SQLite: embedded applications, mobile, edge devices, and local development. It supports a strict superset of SQLite's SQL dialect plus vector search, BM25 full-text, and geospatial indexing — capabilities SQLite requires third-party extensions for.

SQLite is ~600 KB and has a larger ecosystem of existing integrations. Absolute DB's embedded binary is ~154 KB but ships with substantially richer built-in capabilities. Both are zero-dependency, single-file databases suitable for production embedded use.

How does it compare to DuckDB?

DuckDB is an excellent analytical database with a ~12 MB embedded binary. Absolute DB's embedded binary is ~154 KB and covers both OLTP and OLAP workloads. DuckDB is optimised for analytics only; Absolute DB is an HTAP (Hybrid Transactional/Analytical Processing) database with full ACID transactions, concurrent write support, and client-server mode.

If your workload is purely analytical queries over flat files with no concurrent writes, DuckDB is a strong choice. If you need concurrent writes, a client-server protocol, network APIs, or a combination of OLTP + OLAP, Absolute DB is the better fit.

Is it really zero external dependencies?

Yes. Absolute DB links only against libc, libm, libpthread, and libdl (the last only for the WASM JIT on Linux). There is no OpenSSL, no libcurl, no gRPC library, no libarrow, no libparquet, no liburing. Every capability — TLS 1.3, post-quantum cryptography, HTTP/2, gRPC, Arrow IPC, Parquet, Protobuf encoding — is implemented from scratch in pure C11.

This is a hard requirement enforced by the build system. make release will fail if any new external dependency is introduced.

Licensing

What licence is Absolute DB under?

Absolute DB is licensed under the Business Source Licence 1.1 (BSL-1.1), developed by MariaDB Corporation and widely adopted by modern databases (CockroachDB, Singlestore, Citus, etc.). See the Terms page for the full licence text.

What is BSL-1.1?

BSL-1.1 allows free use for non-production purposes and for production use within defined usage limits. After a conversion date (typically 4 years from release), the software converts to an open-source licence (Apache 2.0). It prevents competing cloud providers from offering Absolute DB as a managed service without a commercial agreement, while permitting most other uses freely.

The Community and SME tiers are free (including for production use within their data limits). Professional and Enterprise tiers require a subscription at $49/month and $499/month respectively.

Can I use it commercially for free?

Yes, for most applications. The Community tier (free, no credit card required) allows production use with up to 10 GB of data, 1 tenant, and 10 concurrent connections — suitable for many personal projects, startups, and small applications. The SME tier (also free) allows up to 5 GB of data, 10 tenants, and 500 connections, with HTAP, columnar storage, and Arrow/Parquet export included.

If your application needs more data, more tenants, compliance features (GDPR/HIPAA/PCI), ML serving, or enterprise clustering, a paid tier is required.

Getting Started

What is the difference between absdb, absdb-server, and absdb-lite?

absdb (~154 KB) — the full interactive CLI. Connect to a running server, run SQL queries interactively, manage databases, and run admin commands. Think of it like psql but with more built-in tooling.

absdb-server (~230 KB) — the full network server. Listens on PostgreSQL wire (5433), REST (8080), gRPC (9090), Redis RESP3 (6379), and more. This is what you deploy in production.

absdb-lite (~154 KB) — the embedded library with no network stack. For IoT devices, mobile apps, edge computing, and local-first applications. No server required — embed it directly in your application like SQLite.

What port does it use?

By default, the PostgreSQL wire protocol listens on port 5433 (not 5432) to avoid conflicts with an existing PostgreSQL installation on the same machine. You can change it with --port 5432 at startup. Other ports: REST/GraphQL/WebSocket on 8080, gRPC on 9090, Redis RESP3 on 6379, Raft internal on 9091, C-RAID internal on 9092.

Is it production ready?

Yes. Absolute DB was released on 2026-04-14. All 2,737 tests pass, zero AddressSanitizer/UndefinedBehaviorSanitizer/ThreadSanitizer violations, SQL:2023 100% conformance (150/150 tests). FIPS 203/204/205 KAT (Known Answer Tests) all pass. The codebase has been hardened with AFL++ fuzz testing across all parser and deserialiser paths.

How do I connect existing PostgreSQL tools?

Point your existing PostgreSQL client at port 5433 (or 5432 if you started the server with --port 5432). The following have been verified compatible: psql, pgAdmin 4, DBeaver, TablePlus, DataGrip, psycopg2 (Python), asyncpg (Python), node-postgres, Prisma, Sequelize, Django ORM, Rails ActiveRecord, Spring Data JDBC, Hibernate (JDBC), Npgsql (.NET), and JDBC drivers. No driver changes are needed.

Does it support SQL:2023?

Yes — 100% conformance. Absolute DB passes all 150 SQL:2023 conformance tests, covering UNIQUE NULLS NOT DISTINCT, JSON_TABLE, temporal PERIOD FOR syntax, GENERATED ALWAYS AS IDENTITY, DISTINCT ON, CREATE OR REPLACE VIEW, COPY TO/FROM STDOUT, TRIM ARRAY, and polymorphic table functions. Conformance results are published at absolutedb.com/conformance.

What is the binary size?

Stripped release binaries built with -Os -flto: absdb CLI ~154 KB, absdb-server ~230 KB, absdb-lite ~154 KB. The WASM build (for browser/Cloudflare Workers) is ~3 MB gzipped. For comparison: PostgreSQL's server binary is ~18 MB but requires dozens of shared libraries; DuckDB's embedded binary is ~12 MB.

Does it run on Raspberry Pi?

Yes. Absolute DB runs on any ARM64 target including the Raspberry Pi 4, Raspberry Pi 5 (ARM Cortex-A76), and Raspberry Pi Zero 2W. The embedded absdb-lite binary is designed for devices with as little as 32 MB RAM. On Apple Silicon (M1/M2/M3/M4), NEON SIMD instructions are used automatically, and Apple Silicon is a first-class supported platform with a native Homebrew tap (brew install absdb).

Features

Can it do vector search AND SQL in the same query?

Yes. Absolute DB supports hybrid queries that combine relational JOINs, BM25 full-text matching, HNSW vector similarity search, and graph traversal in a single SQL statement. For example:

SELECT d.id, d.title
FROM documents d
JOIN embeddings e ON e.doc_id = d.id
WHERE d.body MATCH 'neural networks'
ORDER BY e.embedding <=> '[0.1, 0.2, ...]'::vector
LIMIT 10;

This single query uses BM25 to filter relevant documents and then ranks them by vector similarity — a hybrid RAG retrieval pattern — all within the database engine.

How does clustering work?

Absolute DB uses Raft consensus for high availability and automatic failover. A Raft group contains up to 31 nodes (in the current release); one is the leader, the rest are followers. If the leader fails, a new one is elected automatically within seconds. Raft internal communication uses port 9091.

For data distribution, C-RAID handles striping and replication across nodes (see below). The combination of Raft (consensus) + C-RAID (storage distribution) provides both HA and horizontal scale-out.

What is C-RAID?

C-RAID is Absolute DB's distributed storage layer — a RAID-like system implemented over a cluster of nodes using consistent hashing for data placement. It supports RAID-0 (striping for performance), RAID-1 (mirroring for redundancy), and RAID-5 (parity for balanced redundancy and performance). It includes automatic rebalancing, dirty-shutdown self-healing, and predicate pushdown — SQL predicates are executed at the storage node to eliminate unnecessary data movement over the network.

Is it HIPAA / GDPR compliant?

Absolute DB provides the technical controls required for HIPAA, GDPR, PCI-DSS v4, SOC 2 Type II, and ISO 27001:2022 compliance. This includes: automated GDPR erasure (absdb_gdpr_erase()), PHI column tagging, access audit logs, FF3-1 format-preserving tokenisation for PAN data, AES-256-GCM column encryption, SHA-256 hash-chained audit logs in WORM mode, and evidence collection APIs. Compliance is only achieved when you configure these controls correctly — the database provides the tools; your deployment configuration and policies determine actual compliance posture.

What post-quantum cryptography algorithms are supported?

Absolute DB implements all three NIST post-quantum standards published in 2024: ML-KEM-768 (FIPS 203, key encapsulation — replaces RSA/ECDH), ML-DSA-65 (FIPS 204, digital signatures — replaces ECDSA), and SLH-DSA (FIPS 205, hash-based signatures for CA root certs and release signing). For TLS 1.3, a hybrid key exchange combines X25519 with ML-KEM-768 — providing both classical and quantum-resistant security simultaneously. FIPS 140-3 Known Answer Tests (KAT) are included in the build system and pass on every build.

Commercial & Support

How do I get a licence key?

Community and SME tiers require no licence key — download and run immediately, no account required. For Professional ($49/month) and Enterprise ($499/month) tiers, sign up at app.absolutedb.com. A licence key is provided on sign-up and is set via the ABSDB_LICENSE_KEY environment variable or the license_key configuration option. Enterprise licence keys unlock the full feature matrix including zero-trust mTLS, C-RAID clustering, and 100,000-tenant multi-tenancy.

Where can I get support?

Community support: GitHub Issues at github.com/supportcall/AbsoluteDB and the community Discord server (link in the footer of absolutedb.com).

Professional tier: Email support with 2 business day SLA — contact us.

Enterprise tier: Dedicated SLA engineering, direct Slack channel, and 4-hour response SLA for critical issues. Contact enterprise sales for onboarding.

Continue Reading

Quick Start Pricing Troubleshooting

Ready to run Absolute DB?

~154 KB binary  ·  zero external dependencies  ·  2,737 tests passing  ·  SQL:2023 100%

Download Free → View Pricing All Docs