[Gaikai and OnLive are often lumped together under the much-hyped label of "cloud gaming," but the two companies are notably different. Gamasutra speaks with executives from both companies to sort through OnLive and Gaikai's distinct approaches to the emerging space.]
"Cloud computing." Some argue that it's an annoying marketing buzz-term that describes server-based computing that has existed for years, or a way for companies to have ultimate control over their content and their customers. (Go ahead and Google "cloud computing bullshit" and see what comes up.)
Right around 2009, though, we started hearing about "cloud gaming." Two companies emerged out of nowhere that year, both claiming to have figured out how to stream graphically-intensive games from remote servers to lower-powered PCs and even televisions. It would be up to these startups to show that there is some real substance behind the buzz term.
Those two companies were OnLive and Gaikai, and despite being grouped under the "cloud gaming" umbrella, the firms -- led by Steve Perlman at OnLive and David Perry at Gaikai -- are distinctly different beasts.
Both innovators in their respective fields, Perlman is known for his work on QuickTime and web-based services like WebTV, while Perry is known in the game industry as founder of Shiny Entertainment and an active participant in various aspects of the games industry.
The two entrepreneurs are approaching cloud gaming from distinct angles. OnLive is a consumer-facing service that sells streaming game access to consumers via a storefront, giving publishers a cut of those sales.
Gaikai, on the other hand, is business-facing, positioning itself as an "enabling platform" -- retailers and publishers pay Gaikai for its ability to provide streaming technology in order to reach gamers directly.
But the differences go even deeper than the distinct business models. Both companies have very different approaches to utilizing the web, different approaches to their networks and different visions of where exactly the cloud will take gaming. Turns out that there are real businesses, new technologies, and fascinating ideas behind this ambiguous, fluffy term "cloud gaming."
Both Perlman and Perry are probably tired of talking about latency issues with cloud gaming, but the fact is that latency can utterly destroy a gaming experience, which is bad for gamers as well as the publishers who put their games on the cloud.
OnLive and Gaikai have distinct approaches to the latency issue: where OnLive has only three datacenters in the U.S. (Bay Area, Texas, Washington D.C.), while Gaikai currently has 24, with more on the way.
These servers host games that are streamed to gamers' computers all across the country, and data has to move back and forth across long distances.
Perlman defends the strategy of having relatively few datacenters, saying OnLive simply does not need so many. "There's a big misconception about [latency]," he claims. "You see the vast distance, and you assume that's where the latency is. There is much more latency in the last mile to your home on a DSL or cable modem connection than there is in the internet.
"So a thousand miles on the internet, if you have an optimal route, is about, say, 21 milliseconds. But we have DSL connections of 25 milliseconds in the last mile of latency," Perlman explains. "And of course most of the people are closer than a thousand miles; they may be 200 miles or so.
"We'll probably add one more [datacenter] -- we haven't needed one." He says that OnLive has "direct connections" to about 10 different ISPs, adding that 90 percent of OnLive's traffic doesn't go over the "internet," per se, rather directly through the "backbone" of various providers.
Perlman is over using latency numbers in order to justify OnLive. "The reality is that people are using it and they like it," he says. "We could do whatever kinds of scientific measurements we want, but it really comes down to the experience." He adds that a new algorithm is slated to be implemented later this summer that will help further reduce latency on OnLive.
Gaikai's approach to latency is markedly different. Chief strategy officer and former SVP of EA's Global Online Group Nanea Reeves says, "Having the datacenter close to you is going to have the biggest impact on latency. So if you're a company like OnLive, and you're in three or four datacenters... it works great if you're in that neighborhood. The fact that [our datacenters] are distributed is a key differentiator in our speed."
She claims that web performance company Gomez tested Gaikai and OnLive's latency, and Gaikai was faster 25 out of 26 times. A behind closed doors E3 demo of Crysis 2 running off of a remote, high-performance Gaikai server exhibited minimal noticeable latency -- less so than my personal "out in the wild" experience with OnLive.
"It's an architectural decision made early on -- it's difficult to manage, there's a lot of service infrastructure involved in managing that, but it is really what will cause us to win," Reeves adds.
Gaikai head Perry says, "Our objective is to get closer and closer to the users... We're very focused on latency... There's no limit as to how many datacenters we'll have. We'll end up with about 30 by Christmas.
"It's distance. It's the whole transit," he adds. "The most valuable thing besides getting close is peering, and that means going direct... That's been one of our strategic plays, getting great peering with the [ISPs]. That's an enormous part of this equation... That's something that both companies [OnLive and Gaikai] had to solve. There's not a way to get this to work unless you solve that."