Above: A massive crowd gathers in New York City's Central Park to catch 'em all.
A quick glance at the popularity of Pokémon Go makes it seem like a runaway success. More than 7.5 million downloads and $1.6 million in daily revenue. Usage time has already eclipsed WhatsApp, Instagram, and Snapchat. Nintendo’s market cap has risen by $9 billion since the game was released.
And yet, the launch could have been much more successful.
Server outages have plagued Pokémon Go as players overload the service. The problem has become so common and widespread that fans even launched a site to monitor the server status. For a game that had as much lead-up press as Pokemon Go, this is no surprise. Unfortunately, it’s become all too common for games to sputter out of the gates because no one accurately predicted the load at release.
Businesses see large brand and customer adoption fallout every time something like this happens., like with the launches of Blizzard’s Diablo III, Bungie’s Destiny, and almost any other popular game.
The problem is that publishers don’t do the proper sizing, and when demand exceeds predictions, they have no solution to burst resources properly. Multiple bottleneck points compound the problem. In Blizzard’s case, it had congestion issues with the Battle.net authentication servers as well as the game servers.
Why does this happen, and what can companies do about it? It gets a little technical, but every publisher can takes some steps before a game launch to ensure smooth gameplay for fans and a launch that’s successful by every measure.
Why most new games experience server crashes
Above: The desolate landscape of Pokémon Go during server issues.
The launch of a game or expansion almost always causes servers to crash. Many gamers have come to see it as part of the deal. Assuming the limited availability is actually due to poor planning (and isn’t a marketing tactic), why are companies willing to risk frustrations with down- servers that could cause users to drop the game for good?
Part of the problem is that businesses are only willing to invest to reach a certain threshold, and even then, they only conduct rudimentary performance testing as an afterthought. Another factor is that we’ve come to expect server issues around every new game launch, as if it’s normal. This encourages publishers to ignore launch performance issues and ultimately hurts the players and the business.
As applications become more scalable as massively multiplayer online role-playing games (MMORPGs) and more widely available through iOS and Android, this problem is only going to become more pervasive. Only when the examples of failure grow worse or alienate too many users will publishers start to recognize and mitigate these issues proactively.
Looking at the list of games that have experienced server issues around launch, you might ask whether a few bad days prevents a massive success. In the case of Pokémon Go, are brand reputation and revenue even issues given how popular and widespread the game has become and how much money it’s generating?
Absolutely. When faced with servers experiencing constant issues, some players won’t even attempt to download the game due to the news about oversubscription and playability issues. Some won’t download or play because the system is not available. Some will download and try to play the game and quit after a brief period due to playability issues and move on to the next game.
Eventually the system will stabilize as the business adds more resources to correct the availability concerns, but all the trouble prevents a significant population of gamers from becoming subscribers and active players. Even after the issues are corrected, the problems and news around them will keep some players away.
How to prepare for the most successful game launch
Several steps help ensure a successful and outage-free launch:
- Do the research. Publishers should figure out proper sizing by marketing and initial interest based on surveys, feedback, testing, and other measures. Will there be 10,000; 100,000; 1 million; or more than 10 million players in the first weeks of launch?
- Do the testing. Test the application in real-world scenarios (developer, alpha, closed beta, etc.) to determine the performance of the application under all potential conditions. This includes mobile, tablet, Wi-Fi, 3G/4G, hardware specs, and so on.
- Build the infrastructure. Once you know performance metrics, build the infrastructure to support the initial release population. How many servers and how much bandwidth is necessary to support the minimum projections?
- Expect the unexpected. Have a plan in place to adjust the resources available if the demand meets the top end of projections or even exceeds it.
Cloud bursting capabilities can add resources on demand and make other resources available on standby. When we talk about cloud and virtualization, elasticity is one of the fundamental benefits since publishers can spin up (or down) resources to meet application demand.
This is ideal because an application will always be available and use resources only as needed. When not in use, those resources (compute, network, storage) can go to other tasks, as long as the pool of resources is large enough to meet all expected demands.
What will the next game launch look like?
A stable and fulfilling user experience is key for Nintendo to move forward, correct the issues, and make Pokémon Go even more successful as it moves out of the honeymoon stage.
Key metrics include the response time and availability of the application. Maintaining a resilient and responsive gaming system will require continuous monitoring of the end-to-end application response time and adjusting resources accordingly.
The resources needed may be geography-based, an application subset (authentication, rendering, GPS mapping, etc.), or even connectivity-related if there are service provider relationships (since one might have more servers, for example).
By many measures, Pokémon Go is already a cultural phenomenon. Yet it fell victim to the same mistakes many game publishers make during a launch. Players deserve better, and with a more proactive approach to load testing, publishers will see more successful launches moving forward.
Frank Yue is the director of solution marketing, application delivery for Radware and has over 20 years of experience building large-scale networks and working with high performance application technologies including deep packet inspection, network security, and application delivery.