Scale Out the Stateless Tier Without Ignoring State
Horizontal scaling of a stateless tier stops at the first shared dependency, such as a database, cache, or rate-limited API. This interactive lesson finds hidden state and shared bottlenecks in a storefront and names the first limiter. Covers: In-memory sessions, local-disk uploads, long-lived connections, lock contention, hot keys, cache stampedes, third-party rate limits, backpressure. Practice: Stateless scale-out bottlenecks. Prerequisites: Stateless services and load balancing.
A seasonal storefront can add containers during a traffic spike, but that does not make every request path scalable. In a design interview, the dimension being assessed is whether we can locate the boundary between compute that scales independently and dependencies that impose a shared limit.
Stateless compute and request ownership
Statelessness means any healthy web or application programming interface (API) instance can handle any request without needing local-only data for correctness. For the storefront, a load balancer can send GET /products or POST /cart/items ...