Conflating a Queue

Building a memory-efficient queue for high-frequency market data

This easy C++ coding problem tests your ability to design a custom data structure that solves a real operational constraint in market-data systems. Quant trading firms deal with enormous volumes of tick updates; when downstream consumers fall behind, naive queue implementations can exhaust memory. The key insight here is recognizing that only the most recent update for each instrument matters—earlier updates become stale and can be discarded.

The challenge is to implement a queue-like interface that automatically conflates (merges) redundant updates while preserving order for distinct instruments. A strong solution balances simplicity with efficiency, choosing appropriate containers and logic to avoid unnecessary allocations and scans. You'll need to think carefully about what "latest" means in an ordered stream and how to track which instruments have pending updates.

• Data structure selection for fast lookups and insertions
• Order preservation across distinct keys
• Memory efficiency under high-throughput conditions
• Template design and generic container semantics