Cache & Performance

Product Flat Cache

• Global in-memory cache for flattened products
• Concurrent-safe (sync.RWMutex)
• Set PRODUCT_FLAT_CACHE=off to bypass (direct DB)

No N+1

fetchFlatProducts uses GORM Preload with IN clauses — ~10 batch queries regardless of product count.

Benchmarks

• With cache: ~300 req/s, ~1 ms single product
• 100 products, 100 attrs: REST ~25 ms, GraphQL ~30 ms (make test-perf)