How FaaS Transformed Frontend Development at Alibaba’s Taobao

Introduction

This article shares the experience of the Big Taobao technical team in converting their front‑end data‑assembly logic to Function‑as‑a‑Service (FaaS) and answers the three key questions: Why adopt FaaS, what it entails, and how to implement it.

Business Background

Since 2019, the team has been the pioneer of FaaS within the group, applying it to multiple business scenarios such as iFashion and Taobao Sports. The typical data flow of these lightweight applications resides in the BFF layer, where most requirements involve interface and data composition, making them well‑suited for FaaS.

Industry Characteristics

The industry’s data logic is largely built on existing middle‑platform capabilities, with the front‑end performing heavy data orchestration. This "heavy process orchestration, light data assets" nature means the Java side mainly aggregates interfaces, while the front‑end can take over the entire logic, reducing overall manpower and giving front‑end engineers deeper business influence.

Shift in Production Relationship

Traditional front‑end/back‑end separation creates unclear boundaries and high integration costs. By moving the orchestration to the front‑end and letting the back‑end focus on middle‑platform services, a three‑party win is achieved.

FaaS Progress and Benefits

More than ten lightweight applications have been fully FaaS‑ified. Front‑end engineers have evolved from page builders to business partners with ownership of logic, improving overall development efficiency, empowering front‑end, and creating business value.

Solution Architecture

The simplified FaaS architecture consists of three layers:

Serverless Development Platform : Provides a complete FaaS development pipeline from creation to operation, with customizable solutions.

Infrastructure Layer : Offers the serverless runtime, middleware, containers, and high‑availability guarantees.

Underlying Services : Includes common algorithms, data storage, industry‑specific services, and third‑party capabilities accessible via middleware.

Evaluating Suitability

Adoption depends on cost and benefit. Cost factors include increased front‑end workload and technical difficulty (e.g., complex databases, high concurrency). Benefits focus on improved development efficiency and expanded front‑end capabilities.

Migration Strategies

Two approaches are recommended:

All‑at‑once Migration : Rebuild all Java code to equivalent FaaS within a dedicated period. This yields quick results but carries higher risk.

Incremental Migration with New Requirements : Migrate existing code gradually while handling new features, offering controlled risk and workload.

The team prefers the incremental approach.

Stability Assurance

Key concerns include traffic spikes, container concurrency, and fallback mechanisms. Traffic evaluation uses formulas such as: average QPS = daily PV / 50000 Fallback services cache successful responses and serve them from CDN when the upstream fails.

Alibaba Cloud ECS SLA: 99.975%

Low‑Code Exploration

The team built the "Zhilang" platform to visually compose FaaS interfaces, producing over 1,000 APIs for 100+ users. Capabilities include visual orchestration, full development lifecycle, runtime isolation, and an open JSON protocol.

Conclusion

Node FaaS expands front‑end engineers’ capability boundaries, reshaping the production relationship and delivering tangible value. Before starting, teams should assess cost‑benefit, consider traffic, and adopt incremental migration to ensure stability and success.