How to Build a Full‑Cycle DevOps Environment and Embrace Serverless Architecture

DevOps Overview

DevOps integrates development and operations through automation to achieve consistent, repeatable delivery from source code to production.

Key requirements for a Docker‑based DevOps pipeline

Environment consistency : Code built locally must produce identical results in any deployment environment.

Automated code inspection : Every commit triggers static analysis and unit tests to catch defects early.

Continuous integration : The system automatically compiles and packages the code after each commit without manual intervention.

Continuous deployment : After integration the new version is rolled out automatically with zero‑downtime.

Continuous feedback : Inspection, integration or deployment failures are reported instantly to developers and operators.

Rapid rollback : A failed deployment can be reverted to the previous stable version in seconds.

Elastic scaling : Resources are automatically scaled up or down according to traffic load.

Visual operations : A dashboard provides real‑time monitoring of applications, clusters and hardware.

Architecture components

GitLab code repository

Docker container runtime

Jenkins for continuous integration

SonarQube for code quality analysis

Harbor private image registry

Kubernetes for container orchestration

Pipeline workflow

Developers commit code to GitLab.

Jenkins receives a webhook, pulls the latest commit, and triggers the pipeline.

SonarQube runs static analysis and unit tests; failures abort the pipeline and notify the team.

Jenkins builds a Docker image from the source; build errors also abort the pipeline with notification.

The image is pushed to Harbor.

Jenkins triggers a rolling update in the Kubernetes cluster, pulling the new image and replacing pods without service interruption.

Serverless Computing

According to the CNCF Serverless whitepaper, serverless computing lets developers build and run applications without managing servers. Functions are uploaded to a platform that automatically handles execution, scaling and billing based on exact demand.

Alibaba Cloud serverless offerings

Serverless Application Engine (SAE) : A PaaS that abstracts the application layer, supports frameworks such as Spring Cloud, Apache Dubbo and HSF, and accepts deployments via WAR, JAR or container image.

Function Compute (FaaS) : Developers write code locally and upload it with the CLI tools fcli or fun (or via the console). The platform builds a Docker image, stores it in a registry, and runs the function in response to triggers such as OSS events, HTTP requests, CDN, SLS or scheduled jobs.

Function Compute workflow

Write code locally.

Upload the code using fcli, fun or the web console.

The platform builds a Docker image and pushes it to an image registry.

Configure event triggers (OSS, HTTP, CDN, SLS, cron, etc.).

When a trigger fires, the scheduler launches a container from the image, executes the function, and automatically scales the number of containers based on load.

Future outlook (UC Berkeley)

New “BasS” storage services will provide near‑local performance, enabling more workloads to migrate to serverless.

Higher‑level programming abstractions and finer‑grained isolation will improve security and reliability.

Pay‑as‑you‑go pricing will make serverless cheaper than traditional server‑based deployments for most workloads.

Serverless will drive the growth of Backend‑as‑a‑Service (BaaS).

While traditional server‑based computing will persist, its share of cloud workloads will gradually decline.

Serverless is expected to become the default cloud programming model, largely replacing conventional server‑centric architectures.

Conclusion

Because current data‑center utilization is low (≈10 % for online services), large‑scale adoption of serverless can dramatically improve overall resource utilization and reduce costs by scheduling resources on demand. As the model matures, it will reshape programming paradigms, offering both cost efficiency and operational simplicity.