Elastic Auction Room Sharding for Peak Traffic

Problem

Scheduled auctions caused sudden surges where a single room could not handle all users without performance degradation.

Solution

Implemented dynamic room partitioning: users were placed into multiple isolated rooms with separate bid visibility, scaling room count based on load.

Impact

• Sustained peak-time events by distributing load across rooms (no cross-room bid/user visibility).
• Enabled scaling to N rooms based on participant count, improving stability at peak.