Technical Requirements and Architectural Directions for Cloud Databases

Large enterprises such as financial institutions are required to adopt cloud‑native microservice architectures, which in turn demand a cloud‑native foundation software and data platform.

Microservices are service‑oriented, loosely coupled architectures with well‑defined boundaries.

Key characteristics include autonomous services that run independently, interface‑only exposure to applications, and fine‑grained business granularity.

In the cloud‑migration of enterprise architectures, database cloudification is the most critical and challenging part. Database‑as‑a‑Service (dbPaaS) refers to database management or storage systems that support elastic scaling, multi‑tenancy, self‑management, and run on IaaS infrastructure.

According to Gartner, the dbPaaS market will double in the next five years, with 70% of users expected to adopt it, making database architecture evolution a major focus for the next decade.

Technical Requirements of Cloud Databases

Elastic scaling capability: capacity must expand with business demand.

Elastic deployment and on‑demand adjustment: all database functions, not just storage, must be deployable flexibly.

Data reliability and continuous service: always‑online, secure data is a must.

Compute‑storage separation: independent configuration of compute and storage resources.

Multi‑model storage capability: support for structured, semi‑structured, unstructured and graph data.

Self‑management: zero‑downtime maintenance, CI, rolling upgrades.

Self‑monitoring and fault‑repair: reduce operational costs.

Specific scenario support: pluggable components or tools for particular use cases.

Regulatory compliance and security: meet audit requirements and protect data.

Beyond functional improvements, the overall architecture must also evolve.

Cloud Database Architectural Directions

Storage‑SQL separation : the storage engine and SQL engine are loosely coupled. The architecture consists of a storage layer (data management, routing, transaction control, indexing, etc.), an SQL layer (request parsing and dispatch), and a metadata layer.

Multi‑model architecture : a single platform supports multiple storage engines (relational, NewSQL, JSON, graph, object storage, etc.), reducing operational and development costs.

Disaster recovery and multi‑active : the dbPaaS layer should encapsulate multi‑copy synchronization, disaster switch‑over, and high‑availability handling, making these mechanisms transparent to applications.

These directions are illustrated by examples such as MySQL’s plugin storage engine architecture and SequoiaDB 3.0’s MySQL‑compatible, storage‑SQL separated design, which provides elastic storage expansion and multi‑region active‑active capabilities.

Advantages of Cloud Database Architecture

No need for manual sharding: native storage‑SQL separation eliminates complex sharding.

Flexible business support: elastic adjustment of storage and SQL layers.

Multi‑engine compatibility: the same SQL interface can work with various underlying storage engines.

Full compatibility with existing applications: standard SQL parsers ensure zero‑code changes.

Data security and availability: distributed storage with multi‑copy redundancy meets stringent regulatory requirements.

Application Scenarios

Traditional transaction services : high performance, ACID guarantees, and multi‑site deployment for banking transactions.

Historical data services : long‑term data retention, online‑off‑line data integration, and HTAP capabilities.

Real‑time online services : low‑latency streaming, high throughput, and flexible scaling for real‑time asset views.

Image storage services : massive unstructured data, lifecycle management, high reliability, and multi‑model storage.

Conclusion

Cloud databases are a crucial direction for future database development. Their architectures will continue to evolve with demands such as multi‑model engines and compute‑storage separation, and providers like SequoiaDB will keep innovating in cloud‑native designs.

Author: Wang Tao, Co‑founder & CTO of SequoiaDB.