Database Backends for Brick

In this section, we provide an incomplete list of available commercial and open-source databases which support Brick features.

Feature List

• RDF model storage: does the database support the storage of RDF models (this should be yes for all of the below databases)

• SPARQL support: does the database support the execution of SPARQL queries? The relevant standards are:

  • SPARQL 1.1 Query (including property paths)

  • SPARQL 1.1 Update (the ability to write to the database using SPARQL)

• Reasoning/Inference: does the database support reasoning and inference on stored RDF models? Relevant features are:

  • what inference is supported: OWL-DL, OWL-RL, RDFS, OWL-Full, SHACL-AF

  • manual or automatic: is reasoning performed as a manual or batch processing task

• SHACL support: does the database support the validation of RDF models using SHACL shapes?

• Multiple Graphs: does the database support the storage of multiple graphs, and can those graphs be updated/queried independently?

• Scaling and Performance: what are the scaling and performance properties of the database? This includes horizontal scaling capabilities, query performance and storage requirements

Database List

Info

If there is a missing database or our documentation is incomplete, please file an issue

RDFlib + Related Libraries

RDFlib and its related libraries are open-source, production-quality Python libraries for working with RDF data.

Links:

• RDFlib GitHub

• pySHACL GitHub

• OWL-RL GitHub

Features:

• RDF model storage: RDFlib supports the storage of RDF models in several possible backends, including an in-memory store by default

• SPARQL support: RDFlib contains a complete implementation of SPARQL 1.1 (Query and Update).

• Reasoning/Inference: The OWL-RL library supports inference for the RDFS and OWL-RL ontology languages

• SHACL support: pySHACL provides a complete implementation of SHACL and SHACL Advanced Features

• Multiple Graphs: RDFlib is Python library, so multiple graphs can always be managed as separate Python objects. RDFlib supports named graphs (all graph objects can have a Named Graph URI); most RDFlib backends allow graph objects to be linked to a Named Graph URI

• Scaling and Performance: scaling of storage can be provided by the different storage backends supported by RDFlib; additional backends can be added easily. RDFlib’s implementation of SPARQL and OWL-RL are not heavily optimized so performance can suffer on larger graphs.

HodDB

HodDB is an open-source, high-performance, research-quality query processor

Links:

• HodDB GitHub

• HodDB paper (BuildSys 2017)

• Extended HodDB paper (TOSN 2018)

Features:

• RDF model storage: HodDB supports the storage of RDF models on a local filesystem, using BadgerDB

• SPARQL support: HodDB supports a subset of SPARQL 1.1, including property path queries and UNION (specifically not inserts). See the HodDB papers for a more precise list. HodDB also does not support the SPARQL API protocol, and instead requries use of a specific GRPC-based API

• Reasoning/Inference: HodDB supports some basic inference — namely handling OWL inverse and transitive properties.

• SHACL support: HodDB does not support SHACL

• Multiple Graphs: HodDB supports the storage of multiple graphs which can be queried separately

• Scaling and Performance: Due to the restricted query and update model, HodDB provides excellent SPARQL query performance on graphs up to a few hundred thousand nodes. HodDB currently does not support distributed storage

Allegrograph

Allegrograph is a proprietary, commercial graph database supporting RDF and related technologies. Allegrograph also provides a free version with limited features

Links:

• https://franz.com/agraph/allegrograph/

Features:

• RDF model storage: The paid version of Allegograph supports horizontally scalable storage of RDF models; the base version stores RDF models on a single node

• SPARQL support: Allegrograph supports full SPARQL 1.1

• Reasoning/Inference: Allegrograph supports both OWL-RL and RDFS languages

• SHACL support: Allegrograph supports SHACL validation but does not seem to support SHACL advanced features such as inference

• Multiple Graphs: unknown

• Scaling and Performance: Allegrograph demonstrates good performance on graphs typical of Brick

Oxigraph

Apache Jena

Blazegraph

TopBraid