Beyond Even the Cloudless Cloud
A middle layer is needed for low latency to scale
The impending failure of the “Cloud Gaming” movement points toward the end of the cloud decade.
Google’s “Stadia” cloud gaming service is proving a technical failure. The company claimed it would reach hundreds of countries. It reached fewer than two dozen. The company claimed it would solve the game latency problem. It didn’t.
It turns out that for a low latency application like gaming to work, content must be cached close to every player. If it’s centralized, then some players have advantages that render the game worthless. Or you can create a delay, making the whole game worthless.
Forced delays happened over a decade ago, in stock trading. A trading platform found that, by putting their data center close to the Hudson River, it could act on orders from Wall Street before centers further away, based on the speed of light.
In gaming, cutting latency is key. Clouds don’t cut latency. Only ideas like Cloudflare’s “Cloudless Cloud,” which Akamai is now copying, can do that. You need an ad-hoc content delivery network serving every player, or else the game isn’t fair.
Most ideas around the metaverse focus on low-latency applications. Markets and office work are being “gamified.” High bandwidth virtual reality applications are being distributed on a mass basis. But it turns out clouds, built on cash flow, can’t serve this new world alone. Cloudless clouds might, but only to the extent that the necessary infrastructure serving them is close to all customers.
It reminds me of something Amazon learned while scaling their store. You can’t put everything in a huge warehouse outside of town. Eventually you need smaller, subsidiary warehouses, closer to town, the kind that take over malls and big boxes. Connect them all, with technology and infrastructure, and you can scale faster services downward. In the future, this “instant access” is all the advantage Amazon has. But it’s something that can be copied, to a great degree, by Walmart and even Kroger.
The lesson? Clouds aren’t enough. Over time, even cloudless clouds aren’t enough. As with Amazon, a middle layer is needed in cloud before low latency can scale. New visions for clouds, and how they work, are necessary before the metaverse can happen.