Overcoming SDUI Performance Bottlenecks: Oliveyoung's Backend Optimization Success Story Using Local Caching

Oliveyoung adopted SDUI(Server-Driven UI) to ensure UI consistency across platforms and enable fast updates. However, during the high-traffic 'Olyoung Sale' period, they experienced a performance bottleneck where API response times ballooned to 733ms and network traffic increased.

To address this, they adopted a dual caching strategy combining Caffeine(Local Cache) and Redis(Remote Cache). The primary cache, Caffeine, maximized response speed within JVM memory, while the secondary cache, Redis, minimized DB load and maintained data consistency across multiple servers.

To ensure both real-time accuracy and stability of data, they employed two invalidation strategies in parallel.

• Backoffice integration: Immediately deletes related Redis cache upon data modification for real-time reflection.
• Periodic batch jobs: Prevents delays caused by cold caches through Pre-warming once a day.

Through this optimization, they built a stable service achieving response speeds of under 1ms even under traffic exceeding 60,000 requests per second.