Scaling an Image Service with Alibaba Cloud: A Step‑by‑Step Architecture Journey

The author reflects on a product called "萌mark" and shares its architecture evolution, noting parallels with their own team's experience using Alibaba Cloud for an image‑based service.

Rapid Implementation Phase

Alibaba Cloud ECS and RDS were used to quickly build the product environment. The application was deployed on ECS and connected to the RDS database. RDS reduces operational effort, provides high reliability, security, and elastic performance, though keeping a self‑managed database as a backup is advisable.

Image Acceleration Phase

As image storage and access demands grew, the team adopted Alibaba Cloud OSS with CDN acceleration, solving image delivery speed and bandwidth consumption issues. They also retained a copy of images on their own server for safety.

Vertical Scaling Phase

Increasing traffic made direct database queries insufficient, so Alibaba Cloud Memcache was introduced to cache hot, rarely‑changed data.

Horizontal Scaling Phase

When a single server reached its performance limit, multiple ECS instances were purchased and Alibaba Cloud SLB load balancing was employed, eliminating the need to configure and maintain a custom load balancer, hiding server IPs, and allowing seamless addition or removal of backend servers.

Overall, leveraging Alibaba Cloud services significantly reduced operational costs and simplified later expansion, making them well‑suited for the rapid growth phase of early‑stage products.