How SequoiaDB’s Multi-Model Architecture Redefines Cloud‑Native Distributed Databases

Distributed database technology has evolved for years, driven by application and business needs, yet challenges such as distributed transactions, full SQL compatibility, and coexistence of data engines have long troubled users.

SequoiaDB, an open‑source financial‑grade distributed database, has spent six years building its core engine from scratch, choosing an architecture and engine design better suited for cloud database scenarios. The company recently completed a C‑round financing led by Jiashi Investment.

Multi‑Model Database Engine

Since 2007, relational databases have added support for semi‑structured data like XML and JSON, and Gartner predicts a multi‑model future where databases must handle structured, semi‑structured, and unstructured data. SequoiaDB is a typical multi‑model database covering all three data types and supporting transactional, image‑storage, and analytical workloads.

Through a compute‑storage separation architecture, SequoiaDB leverages MySQL, SparkSQL, and PostgreSQL parsers while maintaining 100% industry‑standard compatibility, enabling HTAP (Hybrid Transaction/Analytical Processing). It also provides APIs for JSON and supports unstructured data via POSIX‑compatible file systems and S3 interfaces.

The storage layer uses a dual‑engine design, separating large objects from data records, with unified transaction management, cluster control, replication, and session handling, allowing logical and physical isolation of data and sessions for cloud‑era distributed management.

Since version 1.0 (2013) as a pure JSON store, SequoiaDB added object storage in 2.0 (2015) and full NewSQL compatibility in 3.0 (2017), integrating MySQL, PostgreSQL, and SparkSQL.

Compute‑Storage Separation Architecture

Traditional sharding (e.g., MyCat) merely wraps a single database, while true compute‑storage separation decouples SQL parsing from data storage, allowing independent scaling of each layer.

Cloud databases like AWS Aurora and Alibaba PolarDB place MySQL servers on top of distributed storage, using custom SQL engines and interfaces to achieve loose coupling and dynamic scaling.

The design enables users to choose transaction‑oriented SQL parsers (MySQL, PostgreSQL) or analytical engines (SparkSQL) based on workload, achieving orders‑of‑magnitude performance differences while keeping storage unified.

Because compute and storage are fully separated, users can isolate high‑frequency transactional workloads from high‑throughput analytical workloads on different hardware, achieving true multi‑tenant HTAP capabilities.

SequoiaDB’s 3.0 architecture allows the same data to be accessed by different engines; online services can read from two of three replicas while SparkSQL uses the third, ensuring physical isolation of transactional and analytical workloads.

Elastic Scaling

In the cloud era, applications can scale horizontally on demand, but data layers often become bottlenecks. SequoiaDB uses consistent hashing to enable online, transparent scaling without data migration, providing “zero data migration” for large‑scale workloads.

By adding partitions and nodes, the cluster’s storage capacity and compute power can be expanded horizontally.

MySQL Full Compatibility

SequoiaDB provides a MySQL‑compatible storage engine plugin that runs alongside InnoDB, allowing applications and DBAs to use familiar MySQL tools without learning new syntax.

Migration from InnoDB to SequoiaDB is transparent; data partitioning is invisible to the application, and various migration tools (offline, online, real‑time) are available.

Active‑Active and Disaster Recovery

For financial institutions, multi‑site active‑active and disaster recovery are mandatory. SequoiaDB can deploy a single‑replica remote cluster that synchronizes logs from the primary site, enabling near‑zero RTO and RPO through automatic node splitting.

Conclusion

Based on a multi‑model storage engine and industry‑standard compute‑storage separation, SequoiaDB delivers distributed, HTAP‑ready, MySQL‑compatible capabilities that align with the mainstream cloud‑data architecture.

With elastic expansion, multi‑tenant support, and robust disaster recovery, SequoiaDB contributes to the open‑source database ecosystem and advances China’s database software development on a global stage.