My 13-year-old daughter helped break a data center. I’m so proud.
As one of millions of users relentlessly searching for Pokémon while using the new Pokémon GO game, I am giving my daughter some of the credit (blame?) for the server problems that, as of this writing, are giving users a less-than-fulfilling experience with the game.
For those who actively avoid gaming news, Pokémon GO is a mobile game from Nintendo and Niantic Labs that sends users outdoors with their smart phones to try to “catch” different Pokémon characters in an “augmented reality” format generated by the app on the phone. By all accounts, the game has been wildly successful, so much so that the servers supporting the game can’t keep up.
As a result, users are getting shut out. Instead of accumulating Pokémon, all they get is a dreaded error message: “Our servers are experiencing issues. Please come back later.” Or, in my daughter’s case, the app says it’s loading – but never does. A notice on a Niantic web page for known issues offers little solace:
Due to the incredible number of Pokémon GO downloads, some Trainers are experiencing server connectivity issues. Don’t worry, our team is on it!
It’s difficult to determine what, exactly, the problem is. Reporters at Data Center Dynamics are speculating that the game is hosted on Google Cloud Platform, based on job listings it found that spell out Google is looking for developers to “create the server infrastructure to support our hosted AR/Geo platform underpinning projects such as Pokémon GO using Java and Google Cloud.”
Whether or not that’s true, it’s easy to see how Pokémon GO can suffer performance problems. The game requires a constant back and forth of data between the user and the servers supporting the game. That includes location information from dozens if not hundreds or thousands of users in close proximity, messages back to them that prompt the virtual images to pop up on their phones, data on how many Pokémon each one catches and more.
It’s a perfect example of where edge data center solutions would be of great benefit. Here’s how I picture such a setup would work. In the U.S., for example, a series of edge data centers spread around the country could handle the back and forth of location data and such with individual users. They would also handle any outgoing messages going to players as well as gather stats and generally keep score. Only occasionally would they need to send data up to a central data center, and only a subset of what they collect, such as scoring data.
Such a setup would dramatically reduce the latency of each back-and-forth interaction, as explained in this blog post. Instead of data going from my daughter’s phone near Boston to a central data center in, say, Chicago, it would go to a nearby edge data center. Then maybe the app would load on her phone and she’d still be playing, instead of reading a book (which is fine with me, but probably not what Nintendo wants to hear).
In many respects, Pokémon GO mimics an Internet of Things environment, with lots of devices feeding data to a central site. Steven Carlini, who works for Schneider Electric’s Data Center Global Solutions business, touched on this idea in a recent blog post which talks about how augmented reality solutions are coming to the retail business and how edge computing will be required to support them. What he describes is really quite similar to the requirements demonstrated by Pokémon GO.
If edge computing sounds like something your network would benefit from, check out the free Schneider Electric white paper, “The Drivers and Benefits of Edge Computing.” And good luck with those Pokémon.