BigchainDB

BigchainDB is an open-source blockchain database. It is a NoSQL database that has blockchain properties such as immutability, decentralization and owner-controlled asset. Immutability is the property that once the data is stored in BigchainDB, it can’t be modified or erased. Decentralization means there is no single point of failure in the control and storage process of the data. The owner-controlled asset is the property that only the owner of the asset can transfer the asset. The database provides the user with Byzantine Fault Tolerance (BFT) with no more than one-third arbitrary node failures. There are two main layers in BigchainDB, the consensus layer based on Tendermint the and storage layer based on MongoDB.

History

The first version of BigchainDB was published in February 2016. However, there were many issues with the initial version. For example, the system was not Byzantine Fault tolerant, and there was a single point of failure in the underlying storage database and the system supported only one logical database, which means once got the control of the database the adversary could delete the whole database. BigchainDB 2.0 was published in 2018 and solved all the issues above.

Checkpoints

Consistent Blocking

The consensus layer is based on Tendermint. Tendermint uses LevelDB as blockchain storage. LevelDB builds checkpoints in a blocking manner, see [LevelDB](https://dbdb.io/db/leveldb) for more details. The storage layer is based on MongoDB. MongoDB builds consisten checkpoints, see [MongoDB](https://dbdb.io/db/mongodb) for more details.

Concurrency Control

Two-Phase Locking (Deadlock Prevention) Optimistic Concurrency Control (OCC)

The consensus layer is based on Tendermint. Tendermint uses LevelDB as blockchain storage. LevelDB uses Two-Phase Locking with Deadlock Prevention, see [LevelDB](https://dbdb.io/db/leveldb) for more details. The storage layer on each node is based on MongoDB. MongoDB uses Two-Phase Locking with Deadlock Prevention and Optimistic Concurrency Control, see [MongoDB](https://dbdb.io/db/mongodb) for more details.

Data Model

Key/Value Document / XML

The consensus layer is based on LevelDB, see [LevelDB](https://dbdb.io/db/leveldb) for more details. The storage layer is based on MongoDB, see [MongoDB](https://dbdb.io/db/mongodb) for more details. The main data model is Document/XML.

Indexes

B+Tree Skip List Log-Structured Merge Tree

The consensus layer is based on LevelDB, see [LevelDB](https://dbdb.io/db/leveldb) for more details. The storage layer is based on MongoDB, see [MongoDB](https://dbdb.io/db/mongodb) for more details. The B+Tree is the index for main storage.

Isolation Levels

Read Uncommitted Snapshot Isolation

The consensus layer is based on LevelDB, see [LevelDB](https://dbdb.io/db/leveldb) for more details. The storage layer is based on MongoDB, see [MongoDB](https://dbdb.io/db/mongodb) for more details. LevelDB guarantees snapshot isolation for blockchain storage. MongoDB guarantees read uncommitted isolation level for main storage.

Logging

Logical Logging Physiological Logging

The consensus layer is based on LevelDB, see [LevelDB](https://dbdb.io/db/leveldb) for more details. The storage layer is based on MongoDB, see [MongoDB](https://dbdb.io/db/mongodb) for more details. MongoDB uses physiological logging, LevelDB uses logical logging.

Query Compilation

JIT Compilation

The consensus layer is based on LevelDB, see [LevelDB](https://dbdb.io/db/leveldb) for more details. The storage layer is based on MongoDB, see [MongoDB](https://dbdb.io/db/mongodb) for more details. MongoDB supports JIT Compilation, but there is not enough information about LevelDB.

Query Execution

Tuple-at-a-Time Model

The consensus layer is based on LevelDB, see [LevelDB](https://dbdb.io/db/leveldb) for more details. The storage layer is based on MongoDB, see [MongoDB](https://dbdb.io/db/mongodb) for more details.

Query Interface

HTTP / REST

BigchainDB officially provides the users with multiple programming languages including Python, JavaScript/Node.js, and Java. And there are also many community driven libraries and tools available including Haskell transaction builder, Golang and Ruby driver.

Storage Architecture

Disk-oriented

The consensus layer is based on LevelDB, see [LevelDB](https://dbdb.io/db/leveldb) for more details. The storage layer is based on MongoDB, see [MongoDB](https://dbdb.io/db/mongodb) for more details. MongoDB supports both in-memory and disk-oriented storage, while LevelDB only supports disk-oriented storage.

Storage Model

N-ary Storage Model (Row/Record) Custom

The consensus layer is based on LevelDB, see [LevelDB](https://dbdb.io/db/leveldb) for more details. The storage layer is based on MongoDB, see [MongoDB](https://dbdb.io/db/mongodb) for more details. MongoDB supports custom storage model. LevelDB uses N-ary Storage Model.

Storage Organization

Log-structured

The consensus layer is based on LevelDB, see [LevelDB](https://dbdb.io/db/leveldb) for more details. The storage layer is based on MongoDB, see [MongoDB](https://dbdb.io/db/mongodb) for more details. There is not enough information about MongoDB. LevelDB uses log-structured storage organization.

System Architecture

Shared-Nothing

BigchainDB 2.0 has two layers, the consensus layer, and the storage layer. Each node runs its own MongoDB database as the storage layer and uses Tendermint consensus protocol to communicate with each other. Each node maintains its own MongoDB database to store the structured data and handle queries, it also maintains a LevelDB key-value store to keep the blockchain and support the consensus protocol. Each transaction is proposed by the proposer in each round, and the changes are not committed to the underlying MongoDB until it is settled on the blockchain. The Tendermint protocol can guarantee the BFT under at most one-third arbitrary node failures.

Views

Virtual Views

The consensus layer is based on LevelDB, see [LevelDB](https://dbdb.io/db/leveldb) for more details. The storage layer is based on MongoDB, see [MongoDB](https://dbdb.io/db/mongodb) for more details. MongoDB supports virtual views. LevelDB does not support views.

BigchainDB Logo
Website

https://www.bigchaindb.com/

Source Code

https://github.com/bigchaindb/bigchaindb

Tech Docs

https://docs.bigchaindb.com/en/latest/

Developer

BigchainDB GmbH

Country of Origin

DE

Start Year

2015

Project Type

Commercial, Open Source

Written in

Python

Supported languages

JavaScript, Python

Licenses

Apache v2