Coinbase System Design Review: Order Matching Engine Performance Data for SWE Interview

FAQ

What specific data structures does Coinbase’s matching engine use for order book management? Coinbase uses a combination of a red-black tree (for price-level aggregation with O(log n) operations) and a custom heap structure for time-priority ordering within price levels. This is distinct from simple priority queues used in generic system design problems and specifically optimized for the 10ms p99 latency requirement.

How does Coinbase handle order matching during a regional outage? Coinbase implements synchronous replication to two secondary regions with a Paxos-based consensus mechanism. The recovery time objective is under 30 seconds with zero recovery point objective—no data loss. The matching engine state is reconstructed from the write-ahead log during failover.

What separates a strong Coinbase matching engine design from a generic distributed systems answer? The distinguishing factors are: in-memory order book (no database on hot path), regulatory compliance requirements (complete audit trail), price-time priority with FIFO tiebreaking, circuit breaker patterns for burst handling, and sequential consistency for cross-region replication. A generic system design answer that omits three or more of these elements consistently triggers a no-hire outcome in Coinbase debriefs.amazon.com/dp/B0GWWJQ2S3).