A scheduled exercise where a team injects faults into a system and practices the response, the operational drill that keeps DR muscles alive.
A game day is a scheduled, structured exercise where a team deliberately introduces a fault (failover the database, kill a region, simulate a Sev-1 page) and practices the response. The team picks a hypothesis, defines the blast radius, runs the experiment, captures what worked and what didn't, and writes action items for the gaps. Game days are typically held quarterly or after major architectural changes.
Failover plans, runbooks, and DR procedures decay between uses, configuration drift, staffing turnover, dependency changes. Game days are the only reliable way to keep the response muscle alive: you find the broken procedure during a planned exercise instead of during a 3am Sev-1. The team that game-days regularly recovers from real incidents in minutes; the team that doesn't recovers in hours.
See the part of the platform that handles game day in production.