Edgegap: Making edge infrastructure easier for gaming, enterprise markets

Edgegap’s solution works via a proprietary user-server decision-making engine called Arbitrium. Arbitrium enables developers to launch new servers and applications via an API within seconds. Currently, the primary market for the company is game developers and studios. Still, a modern application delivery network that combines a highly distributed infrastructure footprint with a quick decision-making and deployment solution can be used in the enterprise market as well.

Edge Industry Review spoke with the company’s CEO, Mathieu Duperré, about the company and how it sees the edge computing opportunity evolving.

EdgeIR: One of edge computing’s core promises is improved latency. Explain latency in gaming and why technologies such as CDN haven’t completely solved the challenge.

Mathieu Duperré (MD) CDN is a one-way street delivering static content or video. Video games call for dynamic interaction. If you take any multiplayer game, they’ll use game servers; think of a game server as a referee. The game server is there to be the man in the middle to decide what really happens.

Latency comes into play in different shapes for the gaming industry; there are really four types of lag for gamers. The first one is the human delay which is impossible to remove. The other ways you lag in a video game are network latency, the devices used, and the game code.

The one we’ll talk about is network latency. A survey we’ve done with 2000 hardcore players in the US and the UK found that not only is lag the number one problem, but a third of those players said that they would stop playing a game if there’s too much latency in a given game. From a studio perspective, they lose 34% of their revenue just because they have latency in their game.

When I launched Edgegap in 2018, I saw edge computing as a way to mitigate that problem.

EdgeIR: So why have game developers had challenges in leveraging edge computing platforms?

MD: Amazon, Google, Azure, they’ve built tools, and you need to figure out what you want to build with them. You need resources to develop and manage those tools that are building something. Also, those tools are made for a centralized environment. Game studios are using “fleets” of game servers that are managed by relatively slow tools; those are starting fleets of servers, which are waiting and standing by for players to connect. That means that if you expect 100,000 players every day, you’re going to start a few thousand servers just standing by so that when the players connect their PlayStation or their phone, they can connect to that standby server. And that works for a centralized environment.

That doesn’t work for two reasons. First, there’s a cost associated with that. It may make sense from a cost perspective if you need to have servers waiting and standing by in four or five locations. But if you look at a hundred or even 1000s locations, that’s not going to work from a cost perspective. It’s also a different story where the tools have 15 minutes to start a server and may prepare everything. Looking at a distributed environment, you don’t have the luxury to do that. You need to have those servers up and running in milliseconds.

EdgeIR: What are some of the challenges your company has faced in building your platform for this problem? And how have you solved those challenges?

MD: Initially, I was going to build a policy engine to leverage the edge. Many ISPs and CDNs asked where you need those edges or said, “Bring me a killer app.” I need the infrastructure to sell it because otherwise, it’s just another cloud provider. We figured we would have to build on our own edge infrastructure.

We took what I call the “Pokeman approach”; we collected providers as Pokémon cards and came up with an architecture allowing us to add new providers within minutes. Today, we have access to 350 different locations around the world.

EdgeIR: Recently, you started quietly marketing your solution to enterprises. What opportunities do you see in the enterprise market?

MD: We are being asked by non-gaming customers about our solutions. We have had an IoT company and car parts manufacturers approach us. We had many other use cases, including a company building robots for remote surgery. As a startup, we’re going very slowly and carefully.

The benefit that those customers or those prospects see at us is we already have a worldwide edge, so they can sell their service product on a global matter without having to deal with ISPs in the UK and ISPs in the US ISPs.

It’s a production environment; they can get their applications everywhere without selecting locations. Most other solutions will ask the developers which one you want to use, then say the server will be $5,000 per month. And if you pick 20 locations, that will not cut it. That’s typically how we get in when [enterprises] see our commercial model. They’re typically quite quick at testing it out because our model is when you have an application running, the meter starts, and when that instance stops, so is the meter. We are seeing good growth in the web space [because of that].

Article Topics

Akamai  |  DevOps  |  edge orchestration  |  edge platform  |  Edgegap  |  gaming  |  latency