System Design · HLD Interview Questions — Easy

Answer: High-Level Design (HLD) defines the overall macroscopic architecture of a software system - outlining the major service boundaries, communication protocols, data persistence layers, caching strategies, and infrastructure components before writing code.

• Key Components: Client applications (Web/Mobile), API Gateways, Load Balancers, Application Microservices, Distributed Caches (Redis), Message Queues (Kafka), Primary/Replica Databases, and Object Storage (S3).
• Real-World Example: In an e-commerce platform, HLD models how the User Service, Product Catalog Service, and Payment Gateway interact via asynchronous message queues rather than defining specific class methods.

---

Answer:

• Monolithic Architecture: A single unified codebase and database.
  • When to choose: Early-stage startups (0 to 1 product-market fit), small teams (<15 engineers), low domain complexity, and workflows requiring strict ACID cross-table joins.
• Microservices Architecture: Decomposed autonomous services owning independent databases.
  • When to choose: Large engineering organizations requiring independent team deployment velocity (Conway's Law), services with drastically different scaling requirements (e.g., 100,000 read QPS vs 10 write QPS), and polyglot technology needs.
• Rule of Thumb: Start with a clean, modular monolith; extract microservices only when team scale and database scaling bottlenecks demand it.

---

Answer:

• Relational Databases (Postgres, MySQL):
  • Strengths: Strict ACID transaction guarantees, structured schemas with foreign key integrity, powerful multi-table JOIN queries.
  • Weaknesses: Horizontal write scaling requires complex sharding; fixed schemas slow down rapid iterations. Best for financial ledgers, billing, and core business transactions.
• NoSQL Databases (MongoDB, Cassandra, DynamoDB):
  • Strengths: Dynamic schemas, native horizontal sharding out of the box, massive write throughput, low-latency key-value lookups.
  • Weaknesses: Lacks multi-row ACID guarantees (BASE model), eventual consistency anomalies, no native foreign key joins. Best for real-time analytics, user sessions, chat feeds, and time-series data.

---

Answer: High Availability (HA) ensures a system remains accessible and operational without downtime, typically measured in "nines" (99.999% = $<5.26$ minutes downtime/year).

• Core Strategies:
  1. Redundancy: Deploy at least $N+1$ active instances across multiple availability zones (AZs) with no Single Point of Failure (SPOF).
  2. Load Balancing & Health Checks: Route traffic away from degraded nodes automatically.
  3. Database Replication & Auto-Failover: Use multi-AZ primary-replica replication with automated consensus failover (e.g., AWS Aurora, Patroni for Postgres).

---

Answer: A Load Balancer (LB) sits between clients and backend server clusters, distributing incoming network requests evenly across healthy instances to optimize resource utilization and prevent server saturation.

• Layers: Operates at Layer 4 (Transport Layer: TCP/UDP routing based on IP/Port, e.g., AWS NLB, HAProxy) or Layer 7 (Application Layer: HTTP header, cookie, and path-based routing, e.g., AWS ALB, NGINX).
• Algorithms: Round Robin, Weighted Round Robin, Least Connections, and IP Hash (for session stickiness).

---