Black Forest Database: The Engine Beneath Everything

Black Forest Database™ (BFDB) is Craxel's multi modal, distributed OLTP engine. It is built for real time indexing and deterministic O(1) retrieval, so performance stays constant however large the data set grows.

Think of Black Forest as an onion. BFDB is the innermost layer, the database itself. Black Forest Knowledge Infrastructure is the platform built around it.

One engine natively supports SQL, graph, time series, spatial and key value data, structured and unstructured, which removes the need for several specialised databases and the duplication between them.

Speak with our team
Four data models, one indexGraphVectorTimeseriesRelationalO(1) multi dimensional indexhierarchy preserving hash · indexed on arrival · parallel ingestLow cost object storagepetabytes kept hot, one copy of the dataStructured and unstructured, no duplication between systems
O(1)
Deterministic retrieval. Performance constant however large the data set grows.
5 in 1
SQL, graph, time series, spatial and key value in one engine.
OLTP
Distributed and transactional, queryable against operational data in real time.

One Engine, Every Data Model

Conventional estates run a separate system for each data type, then pay to duplicate, replicate and reconcile between them.

BFDB treats all data spatially, which allows a single multi dimensional index to serve every modality. One copy of the data answers relational, graph, temporal, spatial and key value access patterns.

Complex joins, traversals and analytics run inside the engine, and because BFDB is OLTP the graph itself can be transactional. This suits mission critical workloads that need massive transactional scale, rapid ingestion and real time analytical queries directly against operational data.

One engine
Modalities supported natively
All of these are served by a single multi dimensional index, with no separate specialised database for each and no duplication between them.
SQL
Graph
Time series
Spatial
Key value
Vector
Structured and unstructured
Edge · Local · Cloud

The O(1) Hashing Algorithm

At the core is a patented hierarchy preserving probabilistic hash, which solves a difficult hash collision problem while retaining order. It is the single innovation behind both deterministic O(1) retrieval and high performance searchable encryption.

Because hierarchy is preserved, BFDB keeps the selectivity that conventional hashes discard. Range, bounding, interval and nearest neighbour queries are supported through SQL, SPARQL and a native API.

Indexing happens on arrival and in parallel rather than through a single entry point, with demonstrated rates of 1 gigabit per second using a pair of commodity cloud servers.

Persistence sits on low cost object storage, so petabyte volumes stay hot without partitioning across a large server cluster. BFDB runs as standalone database services at the edge, locally or in the cloud.

Fine grained access control is enforced inside the engine itself, and High Performance Searchable Encryption allows search and analytics over fully encrypted data without the data layer ever holding the keys.

Computed location, not searchBrute force: scan and filterBlack Forest: O(1) direct access
What BFDB Delivers
A single engine for relationships, similarity, time and space, built for mission critical scale.
Knowledge graph
Graphs and time series together
A native API using semantic triples, the most powerful structure for organising and evaluating relationships.
Petabyte scale graphs with real time data, transactional because they are built on an OLTP engine.
Fast traversal between time series and graph, so activity, events and entities connect in one query.
Foundation for AI
Vector support at the data layer
Vector embeddings indexed at line speed alongside the rest of your data rather than in a separate store.
The foundational data layer that fused retrieval and generative AI workloads are built on in the platform above.
Provenance and attribution are preserved at source, so context delivered upward can be traced.
Security
Searchable encryption at source
The same hierarchy preserving hash that delivers O(1) retrieval also enables high performance searchable encryption.
Search and analytics run over fully encrypted data without the data layer ever holding the keys.
Fine grained access control is enforced inside the engine rather than bolted on above it.
Discuss how Black Forest applies to your organisation
Talk to our team about your data, your environment and the decisions you need to make faster.
Speak with our team