How a Mini-Game Scaled to 100M DAU: Architecture, Ops, and Security Lessons

Architecture Upgrade

When the mini‑game "Sheep..." (羊了个羊) went viral, its original architecture—single load balancer, a few PODs, Redis for hot data and MongoDB for persistence—could only handle 5,000 QPS, causing crashes as DAU surged to over 100 million.

After analysis, the team and Tencent Cloud upgraded the system in three directions: compute scaling with TKE Serverless (automatic vertical and horizontal pod expansion), operational improvement using Cloud Log Service (CLS) for real‑time monitoring and alerting, and security hardening with WAF and high‑defense packages to block malicious bot traffic.

Technical Practice: Scaling, Operations, Security

Scaling was achieved by deploying the game on Tencent Cloud TKE Serverless, enabling automatic scaling of containers based on CPU metrics, which added nearly ten thousand cores within seconds during traffic spikes.

Operations were strengthened by integrating CLS for log collection, metric dashboards, and panic alerts, allowing rapid detection of performance issues and quick remediation.

Security was reinforced by inserting a load‑balancer‑type WAF with bot‑behavior management, protecting the service from massive malicious traffic that previously saturated bandwidth and CPU.

The combined upgrades allowed the game to sustain multiple traffic peaks, keeping the system stable even when daily active users exceeded 100 million, demonstrating the importance of high performance, high availability, and high scalability for rapidly growing mini‑games.