HTTP over TCP by default, latency and timeouts, and when to use WebSockets or SSE for real-time updates

Resource-oriented REST endpoints, pagination, idempotency, and keeping API design proportional in interviews

Choosing entities, keys, relationships, normalization versus denormalization, and matching storage to access patterns

Cache-aside with TTLs, invalidation on writes, and reducing database read pressure without hiding consistency tradeoffs

Picking a shard key, balancing distribution against query locality, and avoiding hotspots as data grows

Distributing keys across nodes while minimizing key movement when servers are added or removed

Reasoning about consistency, availability, and network partitions when distributed systems fail

In-memory cache, rate limiter, lock primitive, and fast ephemeral state when sub-millisecond access matters

Full-text search, inverted indexes, relevance scoring, and eventually consistent secondary indexing

Durable event log, partitions, consumer groups, and stream replay for analytics or async pipelines

Entry point for routing, auth, rate limiting, and aggregating cross-cutting concerns at the edge

Wide-column storage built for high write throughput, tunable consistency, and multi-node availability

Managed key-value and document store optimized around access patterns, partition keys, and predictable scale

Relational default for strong consistency, rich queries, transactions, and mature indexing support

Stream processing engine for event-time windows, stateful computations, and low-latency data pipelines

Coordination service for leader election, configuration, and distributed locks in older distributed stacks