DoorDash Database Scaling Interview: Sharding & Partition

What the question asks

You will be asked to explain and apply database scaling techniques to handle large datasets and high-traffic workloads. The focus is on partitioning (splitting data within a single database instance) and sharding (distributing data across multiple database instances), and on choosing strategies that match query patterns and operational constraints.

Core content and context

Describe horizontal vs vertical partitioning, range-based and hash-based sharding, and when to prefer each. Explain how these decisions affect query efficiency, indexing, cross-partition joins, transaction boundaries, and consistency. Use concrete examples in MySQL/PostgreSQL: e.g., shard users by user_id with consistent hashing, or range-shard orders by date to optimize range scans. Discuss replication, failover, and how sharding interacts with read replicas.

Typical interview flow

1. Clarify requirements and workload (read/write ratio, latency, hot keys).
2. Propose a partitioning/sharding scheme and justify it.
3. Draw schema and query examples, discuss indexing and secondary indexes.
4. Handle edge cases: rebalancing, cross-shard transactions, schema changes, and failure recovery.

Skill signals to show

Demonstrate knowledge of ACID vs eventual consistency, CAP trade-offs, consistent hashing, re-sharding strategies, monitoring, capacity planning, and operational costs. Explain practical mitigations for hotspots and how to migrate data with minimal downtime.

By walking through trade-offs, concrete schema decisions, and operational plans, you show both theoretical understanding and pragmatic engineering judgment.