Backend Engineering & Distributed API Development

• RESTful API design best practices, HTTP status codes, idempotency, and OpenAPI/Swagger specs
• Relational database design in PostgreSQL: Normalization, Foreign keys, ACID transactions, and isolation levels
• Deep dive into database indexing: B-Tree, GIN, Composite indexes, and reading EXPLAIN ANALYZE execution plans

• Redis caching patterns: Cache-Aside, Write-Through, Cache Penetration/Stampede prevention, and distributed locks (Redlock)
• Asynchronous event processing using RabbitMQ and Kafka: Producers, consumers, dead-letter queues (DLQ)
• Microservice inter-communication: Building high-speed gRPC services with Protocol Buffers

• Authentication & Authorization: OAuth 2.0, OpenID Connect, JWT signing, and RBAC middleware
• API Rate Limiting (Token Bucket / Sliding Window algorithms) and DDoS mitigation
• Circuit Breakers (Resilience4j), connection pooling (HikariCP/PgBouncer), and health check endpoints

Q1: How do you prevent the Cache Stampede (Thundering Herd) problem when a hot Redis cache key expires?

To prevent thousands of concurrent requests from hitting the database simultaneously when a key expires: (1) Use Mutex Locking (distributed lock like Redlock so only one worker queries the DB and repopulates the cache while others wait), (2) Implement Probabilistic Early Expiration (XFetch algorithm recalculating the cache before it officially expires), or (3) Use Background Refresh with long TTLs where an asynchronous worker periodically refreshes the cache.