A Comprehensive Overview of Storage Development History, Data Models, and Types

Background

The article consolidates information from Yang Chuanhui’s book on large‑scale distributed storage, the "Big Talk Storage" series, and various online resources to provide beginners with a macro view of storage development.

Storage Development History

Storage has evolved from single‑node to internet‑scale solutions, aiming for low cost, high performance, scalability, and ease of use. Today storage manifests as single‑node, centralized, distributed, cloud, and cloud‑native storage.

Data Classification & Model

All storage forms build data models tailored to specific application scenarios. Data is classified into three main types—block, file, and object—and each type follows a corresponding storage model.

Storage Types

Common storage types:

Block storage (e.g., DAS – Direct‑Attached Storage, SAN – Storage Area Network)

File storage (network‑provided file services)

Object storage (key‑value based, separating data and metadata, accessed via RESTful APIs)

Single‑Node Storage

Provides storage engines that expose file, key‑value, table, or relational models on local disks or SSDs. The storage engine implements CRUD operations and determines system capabilities and performance.

Centralized Storage

Includes additional components such as controllers, disk arrays (JBOD), switches, and management devices, forming a more complex system than single‑node storage.

Distributed Storage

Connects independent storage devices via a network to present a unified storage service. System classifications include distributed file systems, key‑value systems, table systems, and databases. Design principles reference the CAP theorem.

Cloud Storage

Built on distributed storage foundations, cloud storage offers Internet‑based services with added flexibility, typically provided by cloud vendors.

Cloud‑Native Storage

Extends cloud storage to meet the dynamic requirements of cloud‑native ecosystems (public, private, hybrid clouds), supporting scalability, S3 API compatibility, and Kubernetes integration.

Reference Projects

Rook – the first CNCF cloud‑native storage project, integrates file, block, and object storage into Kubernetes clusters, enabling self‑sufficient, portable storage across environments.

MinIO – a high‑performance, software‑defined object storage suite that integrates with Kubernetes for cloud‑native data infrastructure.

Recommended Reading

Links to deep analyses of the Chinese AI industry, AI supply chain, and FPGA research frameworks are provided for further study.

Disclaimer

Reposts must credit the original author and source; any copyright issues can be addressed by contacting the publisher.