Agenda
Mewtwo vs 1000 Trainers: Scaling 'Pokémon GO''s Times Square Raid
Ashwin Vasan (Engineering Manager, Pokémon GO, Scopely)
Pass Type:
Academic Pass, Festival Pass, Game Changer Pass, Indie Pass, Start-Up Pass -
Get your pass now!
Track:
Programming
Audience Level:
Intermediate
Format:
Lecture
Vault Recording: TBD
Audience Level: Intermediate
On July 9, 2026, more than 1,000 Pokémon GO players joined the same local Mega Mewtwo Y raid during a synchronized live event in Times Square. This wasn't a typical traffic spike that could be spread across separate matches. Every player was part of one authoritative battle, sharing the same timeline and racing against the same highly visible deadline.
This postmortem tells the story of how the Pokémon GO team took an unstable 1,000+ player battle and built a system capable of handling significantly higher concurrency. We'll walk through the investigation, from realistic player simulations and sudden latency cliffs to overloaded lobby entities, blocked request threads, replay persistence, client rendering bottlenecks, and coordinated recovery tools.
This postmortem tells the story of how the Pokémon GO team took an unstable 1,000+ player battle and built a system capable of handling significantly higher concurrency. We'll walk through the investigation, from realistic player simulations and sudden latency cliffs to overloaded lobby entities, blocked request threads, replay persistence, client rendering bottlenecks, and coordinated recovery tools.
Takeaway
Attendees will learn how to build realistic multiplayer load tests, uncover hidden concurrency cliffs, and prepare high-stakes live events for failure. The talk connects player behavior to system bottlenecks and shows how isolation, bounded concurrency, graceful client degradation, rehearsed runbooks, and carefully scoped operator controls help keep live gameplay reliable.
Intended Audience
This session is for multiplayer backend and online-services engineers, client performance engineers, SREs, and technical leads who build or operate large-scale, high-stakes live gameplay. Familiarity with distributed systems is helpful, but not required. The architecture, traffic patterns, and failure modes will be introduced through clear visual examples.