Scaling Sheep, Sheep, Sheep to Support 100 Million Daily Active Users: A Tencent Cloud Case Study

The article describes how the casual game 《羊了个羊》 (Sheep, Sheep, Sheep) experienced explosive growth, reaching over 100 million daily active users (DAU) within seven days of launch. Its initial architecture could only handle 5,000 QPS, far below the actual traffic, causing service instability.

To address the challenges, the development team partnered with Tencent Cloud and implemented a series of optimizations across three core areas: scaling, operations, and security.

For scaling, they adopted TKE Serverless, a serverless Kubernetes offering, which provided automatic vertical and horizontal pod scaling based on CPU metrics, enabling the system to add tens of thousands of container cores within seconds as traffic surged.

In the operations domain, they integrated Tencent Cloud Log Service (CLS) to monitor API stability, detect exception trends, and perform real‑time panic analysis and alerting, allowing rapid identification and resolution of performance bottlenecks without additional manual effort.

For security, they deployed WAF combined with Anti‑DDoS packages to mitigate malicious bot traffic that was inflating QPS and bandwidth, and also used CDN to offload static assets, multiple load‑balancer entrances for high availability and rate limiting, and transformed MongoDB to a read‑write separation model while using Redis for hot data caching.

After these upgrades, the new architecture withstood repeated traffic peaks and remained stable even after DAU exceeded 100 million, demonstrating high performance, high availability, and high scalability.

The article concludes with three guiding principles for small‑game design: high performance (handling burst traffic with acceptable latency), high availability (overload protection keeping faults within tolerable bounds), and high scalability (horizontal and vertical scaling to balance cost and availability).