regen-network/rhizome-db

RhizomeDB

This repo consists of two related projects which are both at an early, experimental stage:

• rhizome-trees: a library of persistent merklizable data structures
• rhizome-db: a peer to peer database based on these data structures

rhizome-trees

A set of data structures which share the following properties:

• persistent like Clojure collections so that historical state can be maintained with arbitrary branching
• serializable to a variety of storage layers such as key values as well as directly to disk
• support merkle hashes and merkle proofs
• support selective pruning of historical state

These data structures are intended to support a variety of applications which can benefit from such data structures including:

• distributed ledgers (blockchains)
• peer to peer databases
• any database which needs historical state

All the trees will be benchmarked comprehensively to allow for performance comparisons based on different use cases and configurations. As a near term goal, there is the hope that this can help improve the performance of Cosmos SDK based projects.

rhizome-db

WARNING: This is an experimental project. We make no guarantees it will actually turn into anything useful. Maybe it will, maybe it won't. Maybe some other project will achieve the same goals in another way in the meantime, we don't know.

A peer to peer databases which aims to eventually have the following features:

• can be run on any platform (server, mobile, web browser)
• syncs data between peers using CRDTs with eventually consistency
• has fine-grained read/write privacy and selective syncing strategies
• can also work with a light client (i.e. no local data just server connection)
• optional historical state and forking/branching like git, with pruning support
• merkle proofs for using the database at specific points in time as an immutable record
• SQL and GraphQL queries
• an RDF projection with SPARQL queries
• full transaction history with digital signatures
• authorization via various DID methods
• advanced CRDTs for collaborative editing (like Google Docs)
• geospatial and full text indexes
• should be as easy to use as any regular db people use like postgres, mongo, or sqlite
• good CPU, network, and storage performance