Overview of the 58 Faas Function Compute Cloud Platform and Its Core Working Principles

Background

With the emergence of AWS Lambda, the term "serverless" and the concept of Functions as a Service (FaaS) entered the stage, prompting major cloud providers worldwide to launch their own serverless platforms. In 2017, China’s leading cloud providers Alibaba Cloud and Tencent Cloud released their serverless products.

Serverless architecture, literally "no‑server" architecture, does not eliminate physical servers; instead, it abstracts away the need for developers to manage specific servers or hardware, as the hosting environment is provided by the serverless platform.

FaaS, a subset of serverless, is the core service that enables full serverless application development. Unlike traditional architectures, FaaS runs code in stateless containers triggered by events, managed entirely by third‑party services, allowing developers to focus solely on their code.

Function Compute Cloud Platform

The 58‑developed Function Compute (FaaS) platform lets developers write core business logic using supported languages on provided function templates, without needing to provision or manage servers. The platform manages underlying resources such as servers, CPU, memory, and network.

Key Characteristics of the Faas Platform

No Operations Required Fully managed service frees users from infrastructure management, greatly reducing architectural complexity.

Elastic Scaling Provides millisecond‑level real‑time scaling and dynamic load balancing to handle traffic spikes automatically.

Continuous Expansion Event‑driven mechanisms automatically expand applications in response to triggers.

Simple and Easy to Use Developers only need to write core code, enabling rapid function development, deployment, and verification.

Provided Features

Function Management One‑stop operations for writing, building, deploying, taking offline, and rolling back functions directly from the UI.

Log Management, Monitoring, and Alerts Offers comprehensive runtime logs, metrics monitoring, and alerting for key function indicators.

Supported Languages Currently supports Java and Node.js, with plans to add Go, Python, PHP, and more.

Triggers Supports HTTP, WMB message, and scheduled task triggers.

Application Scenarios

Rapid Development of New Business Write business code in a function, build and deploy directly on Faas, ideal for early‑stage product validation.

Simple Feature Development Instead of building a full system for simple features like data queries, developers can implement them with a few functions on Faas.

Event‑Driven Processing Faas integrates event sources such as WMB messages and scheduled tasks, and can be extended to handle file storage events for tasks like image processing.

System Architecture Overview

K8s Cluster Faas runs on a Kubernetes cluster that manages function containers and provides a unified API service.

Function Docker Container Each function runs in its own Docker container launched by Faas on the K8s cluster, with an access component forwarding external requests to the function.

Faas Gateway Acts as the unified entry point, offering management APIs for deployment, scaling, and status queries, as well as an API gateway for function invocation.

Faas Management Platform Provides a UI for creating applications and functions, handling editing, building, deployment, rollback, trigger configuration, monitoring, alerts, and logs.

Triggers Bridge external systems and Faas functions, currently supporting API, scheduled, and WMB triggers.

BaaS Services Backend stateful services such as databases and file storage can be accessed by functions via SCF, API, etc.

Core Working Principles

Function Build and Startup

1. Developers write a function in the management platform, build a function image, and upload it to an image repository.

2. Deploying the function triggers the Faas Gateway deployment API, sending deployment information.

3. The Gateway calls the Kubernetes API to create a deployment.

4. The K8s cluster pulls the image, starts the function container, and awaits requests.

Function Invocation Process

1. An external trigger receives an event and invokes the corresponding function.

2. The trigger sends the request to the Faas Gateway.

3. The Gateway forwards the request to the appropriate function container via its HTTP interface.

4. The function processes the request and returns the result to the Gateway, which then returns it to the trigger.

Function Scaling (Expansion and Shrinkage)

1. The Faas Gateway receives function requests and forwards them to the function.

2. Faas Monitor monitors request traffic via the Gateway.

3. Based on traffic, the Monitor decides whether to scale the function containers up or down.

4. The Monitor invokes the K8s API to adjust container replicas.

Faas keeps a minimum of one container replica to reduce cold‑start latency; scaling occurs automatically up to a configured maximum.

Mini‑Program Cloud Integration

Major providers (Tencent, Alibaba, Baidu, Toutiao) have launched mini‑program platforms, creating new traffic and business opportunities. The 58 mini‑program cloud leverages the Faas platform to provide rapid development capabilities.

Key components offered to mini‑programs include:

Database Service Component : Provides NoSQL storage with automatic creation of databases and tables, supporting CRUD operations.

WOS File Storage Component : Enables file upload and access without requiring separate WOS resource provisioning.

Cloud Functions : Backend functionalities implemented as Faas functions, offering full lifecycle management and HTTP access, and can call other backend services.

Conclusion

The 58 Faas platform is a self‑developed, one‑stop function compute solution with multiple applications already integrated. This article covered its features, architecture, core implementation methods, and the mini‑program cloud solution built on top of Faas. Further details on logging, trigger implementations, and language‑specific features will be explored in subsequent articles.