Understanding Serverless: Evolution, Definitions, and Its Role in Cloud‑Native Architecture

1. Background

Serverless was first proposed in 2012 and became a commercial reality with AWS Lambda in 2014; today it is about six years old. Most developers associate Serverless with Lambda, pay‑per‑invocation, and event‑driven execution. The early hype made Serverless feel like a mysterious "purple‑sweet potato" technology.

Because of AWS's strong influence, many still equate Serverless with Lambda or Function Compute (FC), which is a legacy perception.

In today’s digital transformation, enterprises seek cloud‑native foundations to capture technical dividends; Serverless, being cloud‑native from day one, deserves a systematic review for developers, architects, SREs, and CTOs.

2. Definition

The industry has varied definitions (e.g., CNCF: NoOps and Pay‑as‑You‑Run; Berkeley: Serverless = FaaS + BaaS). In practice, Serverless simply means "Server‑less": developers no longer need to manage servers, contrasting with the IaaS era where one had to provision and maintain servers.

Serverless abstracts away server concerns, delivering intelligent elasticity, rapid delivery, and lower cost.

3. PaaS Rebirth in the Serverless Era

PaaS sits between IaaS and SaaS. Early examples include Google App Engine (GAE) in 2008, which initially impressed but soon revealed many limitations, pushing developers back to IaaS.

Later PaaS products like Cloud Foundry offered more practicality but still required server maintenance, updates, scaling, and capacity planning.

With container maturity and Serverless concepts converging, modern PaaS products combine rapid delivery with Serverless traits (intelligent elasticity, lower cost). Alibaba Cloud’s Serverless App Engine (SAE) exemplifies this blend.

SAE is an application‑focused PaaS that feels natural to developers (deployment, restart, gray release, environment variables, configuration management). It is fully Serverless: no server provisioning, minute‑level billing, and elastic scaling.

Docker‑based containers allow SAE to run arbitrary language applications, solving classic PaaS constraints.

According to the 2020 China Cloud‑Native User Research Report, such Serverless‑PaaS products are increasingly adopted.

Microservices and Serverless

Some view microservices as the current paradigm and Serverless as the next, mistakenly thinking Serverless will replace microservices. In reality, Serverless is a philosophy of not managing servers and can coexist with microservices.

Alibaba Cloud’s SAE integrates microservice capabilities via MSE, offering service registry, governance, and more, allowing developers to continue using microservice architectures on a Serverless platform.

4. Function Compute (FC)

FC, a mature Serverless product, is widely used in front‑end Serverless, multimedia processing, AI, event‑driven scenarios, and IoT. Companies such as Century Lianhua have built entire businesses on FC.

Early limitations (disk size, package size, runtime, memory) have been mitigated by performance instances and pre‑warmed instances to address cold‑start issues.

Typical FC Scenarios

Front‑end Serverless (SFF) – developers write a few functions instead of managing BFF layers.

Server‑Side Rendering (SSR) – Serverless enables efficient SSR for SEO and fast first‑paint.

Multimedia Processing – large video batches can be processed quickly without building complex pipelines.

AI Serverless – Model serving for data‑science workloads benefits from Serverless’s elasticity and zero‑ops.

5. Serverless Containers – ASK

ASK (Alibaba Cloud Serverless Kubernetes) provides a fully managed Kubernetes experience: no node provisioning, automatic scaling, and pay‑as‑you‑go based on CPU/memory requests.

Typical customers include Weibo (rapid scaling for events), Megvii (AI platform), and Qutoutiao (Serverless big‑data computing).

6. BaaS

Beyond compute, Backend‑as‑a‑Service (BaaS) offers API‑driven storage, middleware, and other services that are also Serverless. Alibaba Cloud OSS is a classic example, providing RESTful APIs for universal data storage.

7. Conclusion

Serverless spans front‑end, back‑end, containers, and BaaS, and even includes services like CDN. While not merely FaaS + BaaS, Serverless will dominate cloud computing by lowering costs and boosting developer productivity.

Serverless is a mindset rather than a single technology; when most cloud products become Serverless, cloud computing itself will be Serverless.

Author Bio: Chen Tao, 10 years of software development, 4 years of entrepreneurship, former roles at Taobao and Didi, focuses on cloud‑native, microservices, and Serverless; currently designs and develops the Serverless Application Engine (SAE) at Alibaba Cloud.