Serverless Computing and Function-as-a-Service: Concepts, Benefits, Use Cases, and JD’s Fibonacci Platform

Serverless Computing (Serverless Computing) is an emerging cloud‑computing model that is rapidly growing, with Function-as-a-Service (FaaS) becoming a hot technology.

Function-as-a-Service Overview

FaaS is an event‑driven, fully managed compute service. It decouples business applications into fine‑grained functions, where each function instance is the basic unit of deployment and execution. Functions are triggered by external events and automatically scale elastically based on load, allowing developers to focus on code without managing servers or infrastructure. Its advantages include no server management, rapid development and deployment, event‑driven execution, and fine‑grained resource elasticity.

Advantages of Function-as-a-Service

No server management – users only write function code, set conditions, and publish; the platform handles provisioning and execution.

Event‑driven execution – functions run in response to HTTP calls, message queues, timers, storage events, etc.

Resource elastic scaling – the platform adjusts instance count based on request volume, providing transparent scaling without deployment delays.

Convenient development – developers focus on business logic, using templates and triggers; deployment reduces to uploading code.

Applicable Scenarios for Function-as-a-Service

FaaS is ideal for workloads with fluctuating load, stateless characteristics, and non‑latency‑critical requirements. Typical use cases include:

Web backend services – APIs invoke functions for data access, business logic, third‑party calls, or AI services without deploying separate applications.

Big Data and AI processing – e.g., image processing triggered by storage events, or chatbot services where voice data triggers speech‑to‑text APIs.

Scheduled tasks – timer‑triggered functions handle periodic jobs such as data processing, reporting, or batch operations, allocating resources only when needed.

IoT backend services – low‑frequency, uneven IoT workloads can be handled by functions responding to device events for tasks like weather updates or data storage.

JD Function-as-a-Service Platform – Fibonacci

JD’s technology architecture team and JD Silicon Valley R&D Center built the Fibonacci platform on the JDOS container platform to provide an enterprise‑grade, zero‑maintenance, event‑driven FaaS solution. It offers simple, efficient, fully managed compute services with fine‑grained elastic resource management, greatly improving productivity and IT resource utilization.

① Developers write test functions (online/offline) in multiple languages such as Java, Python, Go, NodeJS.

② Developers simply publish or upload the function code.

③ Trigger‑based execution supports MQ, API gateway, and timer services.

④ Elastic scaling is transparent to users; resources are consumed only during execution.

Key Features

Friendly console with function management, version control, online coding, debugging, and monitoring dashboards.

Enhanced monitoring and elastic policies with multi‑dimensional metrics and auto‑scaling rules.

Multiple trigger modes: HTTP calls, queue consumption, scheduled execution, storage events, etc.

GPU container support for machine‑learning workloads.

Multi‑language IDE supporting Java, Python, Node.js, Go, and extensible to other languages with template libraries.

Overall Architecture

The Fibonacci platform consists of a function console, repository, runtime platform, trigger gateway, and monitoring service, providing full FaaS capabilities such as editing, publishing, triggering, elastic scaling, statistics, and template management.

Function Triggers

Triggers handle incoming events, dispatch requests to functions, collect execution data, and can be extended to sources like DNS or logs.

Function Execution

Local listener services respond to trigger requests, convert input/output data, and invoke the appropriate function. Execution can run in separate processes or threads depending on the scenario.

Function Templates

Templates provide the runtime environment and business interfaces needed for function development, standardizing the creation process.

Function Publishing, Updating, and Deployment

Fibonacci offers multiple fast‑publish and update mechanisms:

Highly abstracted templates focus on changed parts, improving publishing efficiency.

Node‑level external directories manage dependencies, reducing image size and avoiding rebuilds.

Container variables allow logic updates without rebuilding images, enabling rapid push updates.

Monitoring and Elastic Scaling

The platform’s monitoring system collects both resource and application metrics, enabling rule‑based dynamic scaling of function capacity.

We welcome interested peers to exchange ideas and collaborate!

Contact Persons:

Tong Xin, [email protected]

Tu Hui, [email protected]

Dai Dongdong, [email protected]