Understanding Serverless Architecture and FaaS: Concepts, Scenarios, and Impact

Serverless architecture describes applications that rely heavily on third‑party cloud services, often used for mobile apps or single‑page web applications, and can be built entirely on services like AWS S3, DynamoDB, Alibaba OSS, or TableStore.

Traditional deployments required dedicated servers for independent logic, but Functions‑as‑a‑Service (FaaS) enables developers to host such code in a serverless manner, allowing function‑level development, execution, and management without infrastructure concerns.

Since its emergence in 2014, major cloud providers (AWS, Google Cloud, Azure, IBM Bluemix) have offered Serverless services, promising 10%‑90% cost reductions and faster deployments, as highlighted by industry leaders.

Alibaba Cloud launched Function Compute in 2017, viewing Serverless as the next step after IaaS to improve resource utilization, lower costs, and handle business volatility more elastically.

Serverless encompasses many cloud services—object storage, file storage, table storage, messaging, and even big‑data platforms like MaxCompute—offering automatic scaling, load balancing, and pay‑per‑use models.

FaaS is a subset of Serverless, characterized by event‑driven, fine‑grained, real‑time elastic scaling without server management, and serves as the core for building fully Serverless applications.

Serverless aligns closely with micro‑services: both emphasize decoupling, and functions can implement micro‑service units, though micro‑service architectures still require sophisticated deployment pipelines and dependency management.

Typical Serverless scenarios include workloads with pronounced peaks and valleys—where resource utilization drops below 30% in idle periods—and event‑driven data processing such as video transcoding, data extraction, and facial recognition.

Adopting Serverless shifts operational responsibilities from low‑level server upkeep to business‑level monitoring, dependency analysis, and alerting, thereby accelerating DevOps practices.

While Serverless may eventually impact traditional PaaS offerings, it can be deployed on public, private, or hybrid clouds; Alibaba Cloud currently offers it on public cloud.

Future directions involve extending Serverless to edge, FPGA, GPU, and other environments, enabling low‑latency, cost‑effective processing for scenarios like remote wind‑farm monitoring.

The article concludes with a Q&A featuring Alibaba Cloud’s Function Compute R&D manager, covering the technology’s origins, core capabilities (resource scheduling, isolation, debugging, CI/CD), pricing transparency, and short‑ and long‑term product goals.