Backend Developer interview prep for Infosys.

The N+1 problem occurs when fetching an entity with lazy-loaded relationships executes 1 initial query for the parent records and then N additional separate queries to fetch the related children: 1) Join Fetch: Use JPQL `JOIN FETCH` (e.g., `SELECT o FROM Order o JOIN FETCH o.items`) to retrieve parent and children in a single SQL query; 2) Entity Graphs: Use `@EntityGraph` annotation on repository methods to declaratively specify which relationships should be eagerly fetched in a single join; 3) Batch Fetching: Configure `@BatchSize(size = 25)` on the relationship mapping, which groups child fetches into batched `IN` queries (reducing N queries to N/25 queries).

1) Circuit Breaker: Implement Resilience4j around external HTTP/REST calls. When failures exceed a threshold (e.g., 50% over 10 calls), the circuit trips open, immediately returning fallback data without stalling caller threads; 2) Timeouts: Set explicit connection and socket read timeouts (e.g., 2000ms) on all HTTP clients to prevent thread pool exhaustion; 3) Asynchronous Messaging: Decouple non-blocking workflows using message brokers (Apache Kafka or RabbitMQ) instead of synchronous REST chaining; 4) Bulkhead Pattern: Segregate thread pools so that issues in an external reporting service do not deplete the pool used for critical transactional requests.