Database types

TODO

Graphbrain currently only provides one implementation of hypergraph database, which is based on a local filesytem-based high-performance key-value store (LevelDB). By convention, we use the .hg extension to identify directories containing hypergraph databases of this type. In the future, we expect other hypergraph database types to be included in the library (for example, fully in-memory hypergraphs for even higher performance at relatively small sizes, or distributed hypergraphs for huge datasets, fault-tolerance, etc.). Graphbrain is an open source project, so contributions from people interested in developing such implementations are very welcome!

SQLite 3

Pros: universal support, native to Python, very fast, concurrent access, single file storage

Cons: not space-efficient, possibly not as fast as LevelDB

LevelDB

Pros: very space-efficient, very fast

Cons: no concurrent access; very hard to install outside of Linux, directory structure not as nice as single file