Snapchat Database Systems Interview: SQL vs NoSQL

This question tests your ability to compare database system types and choose the right datastore for a product use case.

You will be asked to explain differences between relational (SQL) databases like MySQL/PostgreSQL and NoSQL stores (document stores, key-value stores, e.g., MongoDB or Redis). Discuss ACID properties (Atomicity, Consistency, Isolation, Durability) versus BASE (Basically Available, Soft state, Eventual consistency), and how those guarantees affect correctness for features such as monetary transactions, user profile updates, or social feed delivery.

Walk the interviewer through trade-offs: vertical vs horizontal scaling, schema rigidity vs schema-less flexibility, latency and throughput implications, and operational complexity (sharding, replication, backups). Use concrete scenarios: prefer relational DBs for complex transactions and strong consistency (banking, billing), and choose NoSQL for high-volume unstructured data or fast lookups (real-time feeds, caching, telemetry). Mention indexing, query patterns, and data modelling choices (normalized relational schema vs denormalized documents) and when to use hybrid architectures (polyglot persistence).

Typical flow in an interview: clarify requirements and constraints, list candidate datastore options, justify your choice with consistency/availability trade-offs, sketch high-level architecture (sharding/replication/caching), and describe migration or failure handling. You should demonstrate knowledge of consistency models, data modeling, scaling strategies, and operational considerations like backups and monitoring.