What Enterprise DevOps Can Learn From Video GamesMarch 20, 2014
by Matt Weinberger
This week's Game Developers Conference in San Francisco may not seem like the obvious place to glean insights into the world of the enterprise cloud. But think about it: Most modern games essentially function as a service, with the public cloud underpinning the back-end to ensure a seamless online experience -- and some hiccups aside, they've gotten pretty good at meeting the terms of their service level agreements.
Companies like Loggly, the popular cloud log management solution, are also reaching past their enterprise roots and into the game industry to serve this growing market. Which means that Loggly has learned a lot about how much DevOps matters when running at scale.
Loggly's head of marketing and revenue David Ewart says that there are plenty of lessons that enterprises can learn from the gaming industry when it comes to building and maintaining cloud services that make users want to come back. Chief among those lessons: DevOps is absolutely crucial.
Marketing is great, Ewart says, but DevOps is so much closer to the customer, he says. Eight to nine in the evening is the crucial "gaming hour" in the industry, when most people sit down to play. If a gaming service can't handle that demand and experiences low performance or simply goes down, players are that much more likely never to go back. Where marketing is all about handling new customers, DevOps is all about ensuring a consistent, reliable level of performance.
"If your game goes down, the marketing executive doesn't get the call," Ewart says.
It's especially tricky when you're running at scale. In the age of DevOps-enabled continuous delivery, where code can be deployed to production as soon as it's ready, Ewart says, you run the risk of accidentally breaking things much more often.
For instance, a language patch pushed to a game title in France can introduce a glitch that brings performance screeching to a halt. That annoyance becomes a real catastrophe when you're not being vigilant -- on average, your usage and daily revenue charts may look good, but if you're not keeping a drilled-down perspective and monitoring every aspect of your system, you're missing a huge part of the puzzle. An entire region can literally fall off the map if you're not watching the forest and the trees.
That's Loggly's sales pitch, incidentally: Ewart says that Loggly is the "Nyquil" that enables the on-call engineer to get back to bed more quickly by helping nail down and isolate what went wrong and where. In fact, Ewart wrote a blog post on the very topic of the intersection between DevOps in enterprise and video games.
For video games, service consistency becoming even more popular thanks to the popularity of titles like Candy Crush Saga, where players are encouraged to spend real money for virtual items. A buggy, laggy game that can't connect back to the central store doesn't inspire customers to drop hard-earned cash. For any given retail site, it's the same thing: A slow site means you lose the moment of impetus on an impulse purchase.
In the enterprise, even if your customers are internal rather than external, providing a consistent experience is absolutely key. Think of it this way: If you're a CIO trying to onboard your users onto a cloud experience that just doesn't work, you're never going to see the adoption you want. Performance matters, no matter how you slice it. And that means moving quickly and maintaining visibility.
"You have to get people coming back for more," Ewart says.