It didn’t take long after its early July launch for Pokémon Go to become a phenomenon — and even less time after that for players to totally overload the game’s servers, causing the game’s developer Niantic Labs to pause international rollout.
A chart shared on Thursday by Google, which hosts Pokémon Go and its servers from its Google Cloud Platform supercomputing service, shows just how badly Niantic was caught unaware by the mad rush:
That’s right: Pokémon Go got 50 times more traffic than Niantic expected, and 10 times more than their absolute worst case estimate. As players are intimately aware, it made the game unreliable and buggy, especially in the first week or two post-launch.
Amid the mad rush, Niantic CEO John Hanke put a personal cry for assistance in to Google CEO Sundar Pichai, who fast-tracked reinforcements from Google Cloud.
Niantic became the guinea pig for a new system, called Google Customer Reliability Engineering, where Google Cloud employees actually embedded in and helped add capacity and solve problems with Pokémon Go even as the influx of “millions” of new players slowed things down.
The official Google blog entry has more technical detail on how they solved the capacity problems.
“Niantic’s Pokémon GO was an all-hands-on-deck launch that required quick and highly informed decisions across more than a half-dozen teams,” writes Google in that blog entry.
For her part, Google SVP of Cloud Diane Greene says that she’s “proud” of the role that the search giant played in the “unprecedented” popularity of Pokémon Go — and in reaching players, who she calls “all these zombies walking around, into fountains and so forth.”