Applying Serverless Architecture in a Content Middle Platform: Image Production Service Case Study

What is Serverless? Serverless (literally “no server”) is a way of providing backend services on demand. The provider manages the underlying infrastructure, allowing developers to focus on business logic without worrying about servers. Examples include CDN services, which abstract away node management and load balancing.

FaaS and BaaS are two sub‑categories. FaaS (Function as a Service) represents pure function execution without any backend dependencies, while BaaS (Backend as a Service) provides managed services such as databases, object storage, and message queues. In practice, Serverless = FaaS + BaaS.

Why the content middle platform adopts Serverless

Developers’ basic demands: reduce server costs, accelerate iteration and deployment, and avoid HA and scaling concerns.

Content‑platform attributes: a large number of generic services (e.g., image production) serve hundreds of business lines, requiring high availability and dynamic scaling.

Image‑production service attributes: fine‑grained functions such as cropping, scaling, watermarking, QR‑code generation, etc., demand low‑cost, low‑maintenance deployment.

Image Production Application Cases

1. Self‑contained FaaS – A set of 12+ functions (image composition, cropping, QR‑code generation, video frame extraction, etc.) serve internal editors, moving the workflow from offline tools like Photoshop to an online platform.

Developers only need to specify service name, resource quota, and scaling policy; the platform handles deployment.

2. Synchronous FaaS + BaaS – After image processing, the result is uploaded to cloud storage and distributed via CDN. The pipeline (processing → storage → CDN) remains lightweight and fits CMS image‑upload scenarios.

3. Asynchronous Event‑Driven Model – Instead of the business explicitly invoking image processing, a cloud‑storage change event automatically triggers a function that compresses, watermarks, and persists the image. This decouples processing from business logic and handles traffic spikes gracefully.

4. Serverless Workflow – Complex image‑production pipelines (e.g., AI recognition, video frame extraction, image enhancement) are orchestrated using a Serverless workflow engine, eliminating the need for external workflow servers. Benefits include:

Low maintenance cost – no virtual machines to manage.

Simple deployment – no middleware or resource isolation concerns.

Full decoupling – business only provides input parameters.

Flexible configuration – a workflow can be described in fewer than 100 lines, reducing configuration effort by ~90%.

The workflow automatically branches based on input type (raw video, MP4 segment, GIF) and runs serial or parallel image‑processing functions to produce the final output.

Summary and Outlook

Key takeaways: decoupling, asynchronous processing, workflow orchestration, and function composition. Future work will deepen event‑driven integration with cloud storage and expand Serverless workflow adoption across all image‑production services in the platform.

Overall, the content middle platform’s early adoption of Serverless demonstrates significant cost reduction, scalability, and operational simplicity, paving the way for broader cloud‑native transformations.