If you wanted to know what a zombie apocalypse would look like then head to Manhattan, people are literally falling over themselves while they catch Pokémon.
It’s no surprise that the Pokémon Go app has become a social phenomenon given the series’ ever-popular catchphrase – “Gotta catch ’em all!”, but there have been several other side effects that developer Niantic Labs will be less keen to shout about.
Players are being taken over by an irrational fear – fear that the Pokémon they’ve caught won’t end up in their PokéBall because the game will crash while other users have trekked to the location of a rare catch only to find they can’t login to the app at all.
[easy-tweet tweet=”Pokemon Go users have trekked to the location of a rare catch only to find they can’t login to the app” hashtags=”tech”]
Within six days of launching in the US on 6 July, Pokémon Go had been downloaded by 21 million users. While some players were able to chase down a Pikachu or Psyduck straight away, others had to grapple with another monster altogether – downtime.
The app then launched in Europe on 16 July, but again Niantic had not correctly predicted the demand generated by new players. The game was quickly overloaded as servers could not handle authentication and application requests.
It’s clear that plans to scale out the appropriate pieces of infrastructure including registration servers, authentication servers and application servers were unfortunately insufficient. The downtime has had such an effect that hacker groups like PoodleCorp have claimed responsibility, although there is little evidence to conclude an attack caused the outages.
The Google Cloud Platform, which hosts the game, has come under pressure for failing to handle the demand. Even arch rival Amazon Web Services poked fun at Google, with CTO Werner Vogels tweeting ““Dear cool folks at @NianticLabs please let us know if there is anything we can do to help! (I wanted that drowzee)”.
In theory the cloud should offer all the capacity you need, but phenomenal demand has caught both Google and Niantic Labs out. So would it be different for Amazon? Where do ambitious developers turn if they want to capitalise on the latest craze with their own augmented reality app?
[easy-tweet tweet=”In theory the cloud should offer the capacity you need, but phenomenal demand has caught and Niantic Labs out.” hashtags=”cloud, tech”]
In many ways it’s a case of keeping things as simple as possible. Load balancing, which distributes the workload of your systems to multiple individual systems, has come a long way in a relatively short space of time, evolving into application delivery controllers (ADCs) that integrate core load balancing with an array of advanced application services, including security.
The good news is that a simple foundation can be rapidly implemented and then scaled up as necessary. This way, developers can make sure they keep the monsters on the screen, rather than in the system.
If you are considering how best to launch a new application, here are six characteristics of the future proof load balancer:
Keep it simple, keep it smart
Place emphasis on performance, security and adaptability. The trick is to make sure you have foundations in place so that you can easily add functionality down the road.
It has to be customisable
A basic ADC set up will get you up and running, but having an optimised solution that specifically addresses your needs will result in faster application rollout times and increased operational simplicity.
Virtualise
It’s not recommended to use an existing ADC to serve multiple applications. This can impact end user experience at peak times and even during cyber attacks. New virtual ADCs (vADC) isolate any issues with your applications and can be configured differently without affecting neighbouring set ups. vADC is also much faster and easier to launch as the time needed to add a new ADC unit, rack it, connect it to the network and configure is dramatically cut.
Build as you go
Deploy an ADC solution that can grow as the business scales by allowing enhancements to be added via a modular framework.
Meet the application SLA challenge
End users expect the same quality of experience at all times. Make sure your ADC delivers applications consistently and provides the tools to monitor and manage SLAs.
Connect to next-gen switches
Adoption of 10GE and even 40GE ports is taking off. Ensure your ADC can connect to these switches to enable your applications to benefit from the latest technology.
It is unfortunate, though understandable, that Niantic struggled with the launch of Pokémon Go. But the real test will be after the initial surge of interest cools. Will the app continue to struggle? It may be riding high on a wave of nostalgia mixed with new technology for now, but most players will eventually drop away if problems persist.
It would be crazy to suggest that Pokémon Go has been anything else than a tremendous success, but the launch problems have nevertheless dented the developer’s reputation, along with the platform it depends on. Continued issues could directly impact potential revenue due to lost players who either downloaded the game and had problems or decided not to download it at all.
Hacker groups claiming responsibility in order to gain media attention will further magnify any issues. Even if the claims are untrue, it screams out that the infrastructure is vulnerable and encourages further attacks and negative attention to what is otherwise a popular and successful game.
I’m sure that Niantic knew they would have a hit on their hands when they developed Pokémon Go, but even they must be surprised at the magnitude of its popularity across the globe.
I’m equally sure we will see many more innovative uses of augmented reality applications in the future. Niantic has forged a path for others to follow, but highlighted the need for a carefully considered load balancing strategy.
It’s not just good practice – it’s super effective.
Adrian Crawley is the VP of sales for Synack’s EMEA division. With a long career in cyber security, Adrian advises companies on strategy, and in particular how to secure the enterprise and applications using crowdsourced security testing models.