Chubao Database (CBDB): A Second-Generation Elastic Multi-Model Database Service

Product Overview

Chubao Database (CBDB) is a second-generation elastic database service launched by JD's technical architecture. As a multi-model distributed database system, it provides both data capacity elasticity and data model elasticity: one system supports multiple models and is compatible with various database access protocols.

Product Vision

In 2017 we released a first-generation elastic database based on MySQL storage engine and protocol, offering automatic scaling and improved operational efficiency. However, developers typically use MySQL for structured data, MongoDB for semi‑structured documents, ElasticSearch for search and analytics, and Neo4j for graph relationships, leading to a complex technology stack and increased hardware/maintenance costs due to data copying between systems. Building on years of MySQL, NoSQL, and NewSQL experience, we introduced CBDB as a second‑generation elastic database platform that delivers dual expansion of data capacity and data model.

Product Features

Multi‑model Supports KV, Document, Graph, and SQL models; a single engine, single dataset, multiple API access modes.

Multi‑protocol RESTful: data query and write via RESTful protocol. MySQL: binary‑compatible with MySQL protocol, accessible by any MySQL client. ElasticSearch: compatible with ElasticSearch DSL, providing document search service. Gremlin: compatible with Gremlin graph traversal syntax, offering graph processing capability. GraphQL: supports GraphQL query language.

Container‑native All internal modules run in containers on JDOS, scheduled uniformly by the Ark platform.

Horizontal Scaling Service capacity scales linearly with cluster size; supports pre‑sharding and merging.

High Availability & Strong Consistency Implements Paxos replication protocol for cross‑availability‑zone strong consistency, delivering high reliability and availability.

Distributed Transactions Supports cross‑node distributed transactions with repeatable‑read isolation level.

Self‑Healing Supports automatic fault recovery and automatic replica balancing and replenishment.

Roadmap October 2018: v1.0 released, providing Elasticsearch‑compatible document search and Gremlin‑compatible graph database functions. January 2019: v2.0 released, added MySQL‑protocol relational query/write and built‑in RR‑isolation distributed transactions.

Try‑out contact: [email protected] (This article is republished from JD Commerce Technical Architecture (ID: TIGCHAT))