How Huawei Deploys a Cloud in One Day: Building a Digital Highway

Strategic Background

China’s “14th Five‑Year Plan” and the 2035 vision emphasize accelerating digital development and building a “Digital China.” Cloud computing, with its “everything‑as‑a‑service” model, provides on‑demand compute, storage, network, AI, big‑data and governance services, becoming essential infrastructure for the digital economy, society and government.

Because cloud platforms are inherently complex—requiring data‑center construction, hardware procurement, configuration and software deployment—the initial build can be lengthy and may hinder digital transformation timelines.

Huawei’s One‑Day Cloud Goal

Huawei Cloud demonstrates how to replicate a full‑stack cloud platform in a customer’s data center at the speed of “one cloud per day,” effectively constructing a high‑speed digital‑transformation highway.

Complexity of Cloud Platforms

Building a cloud platform is comparable to constructing a large building; the completion time directly impacts when the “mall” (digital services) can open. Huawei Cloud Stack’s logical architecture includes a resource‑pool layer, common components, cloud‑service layer and operation system, each consisting of dozens to hundreds of micro‑services, databases, middleware and load balancers, making its complexity several orders of magnitude higher than a typical ERP system.

Key Measures for Rapid, Low‑Cost Delivery

Product R&D Phase : Adopt a “building‑block” architecture that ensures high cohesion and low coupling for each logical layer and service unit.

One‑Click Automated Delivery Platform : Use a self‑developed platform that drives bottom‑up automated deployment of the entire cloud.

Automated Acceptance Tools : Verify delivery effectiveness and reduce on‑site debugging costs.

Standardized Data‑Center Modeling : Provide data support for future upgrades and ensure long‑term sustainable evolution of delivered sites.

Building‑Block Architecture

The modular approach treats each service as a Lego‑like component with a standard “slot.” Just as prefabricated building parts are assembled on site, Huawei’s architecture enables rapid, controllable construction.

R&D Practices for Deliverability

Use proprietary modeling tools to design and review delivery and deployment models for each service unit.

Apply layered architecture with clearly defined interfaces and API governance to reduce integration complexity.

Employ domain modeling to extract common components as services, improving standardization and delivery efficiency.

Define a unified installation‑deployment framework and standards across teams.

Example: ECS Service Design

During design, the ECS service incorporates deliverability considerations such as detailed deployment models and dependency declarations, illustrated in the following diagram.

Development Phase Automation

Each service provides an automated deployment plugin package that describes every sub‑step, parameters, dependencies and scripts, shielding delivery engineers from internal complexities and offering a uniform delivery interface.

HCSD – Highly Flexible Visual Automated Delivery Platform

The Huawei Cloud Stack Deployment (HCSD) platform acts as an automated “Lego‑building robot.” Engineers import software packages and atomic capability libraries, select required services via a visual UI, provide parameters, and launch the deployment. The engine builds a DAG from plugin dependencies, performs topological sorting, and executes tasks concurrently. After deployment, automated E2E acceptance cases run and generate reports.

End‑to‑End Delivery Flow and Effort

For a 200‑node physical‑machine deployment, the overall software delivery process and man‑hours are shown below.

Automated End‑to‑End Acceptance

Traditional manual acceptance is impractical for a distributed system like Huawei Cloud Stack. By embedding testability into design, E2E test cases are coded and packaged with delivery artifacts. Automated acceptance covers API calls (e.g., VPC creation, VM image upload, VM creation, network verification, VM deletion) and UI interactions with the cloud console, capturing screenshots.

This automation reduces acceptance time from days to hours, generates both web and Word reports, and lowers the skill requirements for delivery personnel.

Standardized Model for Sustainable Upgrades

Huawei Cloud Stack uses ManageOne CMDB to store core operational metadata. Data models are defined via JSON schema, with automated validation on each code commit. During release, services package their operational baseline data, which HCSD writes into the CMDB, enabling one‑click end‑to‑end upgrades and long‑term sustainable operation for enterprise customers.

Conclusion

From architecture design to code development, release and delivery, Huawei Cloud Stack has established a highly standardized pipeline. Future work will continue to innovate delivery models, rapidly bringing new public‑cloud features to customer data centers and building a high‑speed highway for digital transformation.