How Alibaba Cloud Serverless Boosted Efficiency and Cut Costs by 60%

Background

海牙湾 (G‑Town) is a technology company founded in 2006 that provides digital transformation solutions for finance, cosmetics, energy and other industries. It operates a platform built on Alibaba Cloud ECS servers, serving over 16 national joint‑stock banks and handling diverse scenarios such as points redemption, rights platforms, smart e‑commerce, corporate procurement, and gift customization.

Business Challenges

Resource waste and insufficient elasticity : Distributed deployments on ECS instances resulted in CPU utilization often below 20%, with many servers idle, leading to high costs and inability to handle traffic spikes during major events.

Limited observability : Reliance on basic system metrics and keyword log analysis made troubleshooting slow and ineffective.

Low operational efficiency and complex releases : Manual deployments required per‑instance adjustments, scripts became hard to maintain at scale, and release processes were error‑prone and time‑consuming.

Solution Architecture

The technical team selected Alibaba Cloud Serverless Application Engine (SAE) as the core platform, complemented by Cloud Native API Gateway (formerly MSE Gateway), MSE micro‑service governance, and ARMS for monitoring.

1. Serverless Application Engine (SAE)

Elastic scaling : SAE automatically adjusts instance count based on load and can scale down to a single instance during low‑traffic periods, reducing idle resource costs by up to 80% vCPU.

CPU burst capability : Instances can double CPU specifications during startup to eliminate cold‑start latency.

Idle capacity reduction : SAE can shrink to minimal instances, cutting resource waste dramatically.

Java cold‑start optimization : Integration with Dragonwell JDK improves JVM performance, shortening startup time by about 30%.

2. Automated CI/CD and Deployment

Integration of kubectl-sae with existing CI/CD pipelines enables one‑click JAR build and deployment.

YAML files define application specifications (CPU, memory, image path) and Shell scripts handle build and upload.

SAE timestamps each release, supporting instant rollback to previous versions.

Release strategies include batch rollout, canary deployment, and full‑link gray release, ensuring zero‑downtime and rapid fault recovery.

3. Full‑Link Governance and Stability

MSE micro‑service governance : Provides gray‑release, lossless online/offline, and traffic protection features.

High‑availability design : SAE natively supports multi‑AZ deployment with zone‑affinity, enabling automatic traffic switching and fast fault recovery.

Cloud Native API Gateway : Delivers high throughput (tens of thousands of requests per second), DDoS protection, rate limiting, IP black/white lists, and flexible routing policies.

Business Value

Cost efficiency : Resource utilization rose from <20% to >50%, and overall cost savings exceeded 60%.

Operational cost reduction : Automated deployment and monitoring lowered manual effort, improving team efficiency by 40%.

System stability and high availability : Zero‑downtime releases and elastic scaling raised availability from 99.5% to 99.99%.

Development agility : Deployment time shrank from hours to minutes (≈3× faster), and fault‑locating speed improved by 30%.

Future Outlook

The success demonstrates that Serverless combined with cloud‑native technologies is a strategic path for enterprises seeking cost reduction, efficiency gains, and competitive advantage, and serves as a benchmark for digital transformation across industries.