We recently had the pleasure of interviewing the founder of Edgegap, Mathieu Duperré, to learn more about the company's edge computing technology which supposedly reduces lag in online games by up to 58%.
The Montréal based startup closed a $1 million seed financing round in early December to develop its innovative gaming infrastructure; the investments were made by Konvoy Ventures and Hiro Capital, two venture capital funds focused on the gaming industry.
Needless to say, it is a bold claim and one that would interest dozens of millions of gamers globally. Ever since online gaming became a thing, lag has been the cause behind so many fits of anger when the games fail to respond timely to our inputs. To learn more, we set up a lengthy Q&A session with Duperré and below you can find the full, massive transcript.
It is a long read, but also a very interesting one if you're interested in lag-reducing technologies, edge computing, machine learning and so forth.
Hello, Mathieu. Please describe your startup, Edgegap, to our readers.
We're an early stage startup based in Montreal, I founded Edgegap a year or so ago. I'm coming from Cisco Systems and when I was at Cisco, I had discovered that edge computing was going to be big. If you're not familiar with edge computing, edge computing is a new notion in terms of infrastructure where you have microdata services spread across a territory instead of having a very big server farm centralized one. Three years ago at Cisco, I had seen that edge computing was coming and there was going to be big. Being a gamer myself, I thought, I should get those two passions together. And I founded Edgegap because I thought that this was going to be a good way to reduce lag.
Back then I was playing the pirate game, Sea of Thieves from Microsoft, and I was having a lot of lag even though we're in Montreal, it shouldn't happen here. I founded Edgegap last year based on a concept that I came up with, to use this distributed environment and connect tightly with the backend from gaming studios to leverage these infrastructures. So that's the story behind that. On the second week of starting Edgegap, I flew to Berlin to participate to a congress on edge computing. Once there I was asked if I wanted to participate to a hackathon, and if I was participating I didn't have to pay the ticket. So I said, yeah, sure. Why not? And I ended up winning with a very, I mean, it was a prototype with just a bunch of scripts. I ended up winning on the stage and presenting to a bunch of executives. And one thing led to another after winning the prize and everything, folks were coming to me and say 'Hey, can I try this? Can we integrate with you again?' And I didn't hide anything. It was two weeks into the project 'Oh, yeah, sure. Sure. Let me get back to you'. I flew back to Montreal and quickly opened up a small office, hired a bunch of guys here to help me develop the first version of our software.
And that went very quickly because of that success that I had very early in the project. So we did that and kind of released the first version of our solution early in 2019. We started working with a bunch of studios. Since we're disrupting the way they are doing business today, it takes more time to adapt their technology. But once they are there, when they see the result, they're like 'Oh, yeah, this is really cool. This is a good idea and we want to dig into that'. So we've been working with Ubisoft, just to do a proof of concept because Ubisoft as a lot of other studios, they were asking 'This is cool, Mathieu, we understand what you're trying to do. But how much is it going to be better? Is it 5% better? 10% better?' So we did the test with Ubisoft using Far Cry 5, spent a few months back and forth. They've shared data and we did a few rounds of these tests. Their goal was for us to come up with an apple to apple comparison, they already used AWS in this case, and they wanted to say to see how good it was going to be. And I want you to compare the same player because if you compare two different players, it's irrelevant. So we ended up doing the test at the end with the same set of players throughout four days, a few tens of thousands of transactions or matches. And we've shown that Edgegap could reduce lag by 58%.
That's a lot!
It is a lot. And we did that worldwide. In some regions, we could see that lag was reduced even more. Especially if you live far from one of the public cloud infrastructures, that's going to have a bigger impact. Back then when we did those proof of concept we only had, I think, like 65 or 70 locations. Now we have 220 and we're trying to add more and more every week. The more locations we add, the better the decision can be for our Edgegap software. So that's the story behind Edgegap.
Since you are in Montréal, I'm guessing you might have met a lot of interested folks as there are many other studios beyond just Ubisoft.
Very true, very true. In Montréal, we have those tax laws that are very appealing for studios and publishers. So for the last fifteen years, we had those laws. Most studios, most of the large studios have been building or they already have an office here, all the big guys. I think last time I heard we had 500 different studios in Montréal. There are smaller ones, but there's also a lot of big ones.
Indeed, it's a major hub for gaming development. Regarding your work with Ubisoft, though, did they make any decision on whether to use your software in the future yet or is it still an ongoing discussion with them?
It's still an ongoing discussion. Officially it's still an ongoing discussion. We're embarking on a test with them. I mean, I'm going to say a test but more of a certification, if you will. It's a certification that needs to happen before they start using it officially. We're hoping to do a joint press release with Ubisoft in the next few weeks. It's just that with the investment round closed recently, we've been pretty busy, but we're hoping to do a press release with Ubisoft. We'll see how it goes. But the goal is for us to be certified within the Ubisoft environment so that moving forward every production within Ubisoft can start leveraging the technology.
That would be certainly big news for Edgegap. I'm wondering if your solution can be potentially even more useful for smaller developers, by the way. Are you in touch with them as well?
Yeah, you're asking a good question there. Since we're a startup and we're still very early we try to spend most of our time on larger studios upfront because we don't want to spread too wide. The goal though is for us to open up, in a few months from now, a self-service portal so smaller studios can onboard them themselves without having to talk to us. That's part of the plan. We want to open up the service so we can go more public about or catch the smaller studios, but for now, we're focused on triple-A studios, publishers and medium-sized studios with over 300 hundred employees. I don't want to say they are worth our time because every studio is interesting. It's just that we have so very thin of a resource pool for now. I'm hiring, so we're going to have four new guys starting after the holidays. And hopefully, in April we have two more flying in from Paris. And between January and April, I may be able to have more based on what's going to happen with a few projects we have. So it's all about having the right resources. I'd rather not have new customers if I cannot serve them well. So the self-service portal is going to call for bigger infrastructures right now. We're trying to just keep it smaller, but do it well. It's a new technology and we want to get it right the first time.
Makes sense. Regarding the recent investment round, could you go into the specifics of how it will be used to improve your technology at Edgegap?
Yes, yes. It's a tricky question. I see ourselves a bit like, I'm going to say an R&D shop in a lot of ways. Because studios today, they use standby instances that they have this notion of fleet where you have a cluster of standby virtual machines waiting for people to connect. We see live quite differently. We see live with more of an on-demand approach, a dynamic orchestration, and that leads to a lot of new possibilities. So we are starting in January to work with a local university here to investigate more of the whole endeavor within Edgegap. Right now what we're doing is we're working with studios to distribute payload in its computing environment. We are gathering a lot of data as things are happening. The next step, what we want to do in the next 12 months, is to start taking that big data and do machine learning on what happened and tweak the rules or tweak the way we orchestrate those gaming deployments.
Let me give you an example here. So we collect a lot of data with Edgegap, and perhaps we know that every day at 4 pm around a school, I have kids coming out of school, they want to play Fortnite and we have an edge computing location close to that. We maybe should start a few instances of Fortnite around that area, because we know it's going to be needed. Or, when we have players from a certain provider playing against or with players from another provider, in the end, the result is not right. So we see good results initially, but as the game goes by as the match goes through, lag goes through the roof for some reason in between players. So maybe we should tell the matchmaker 'Hey, don't pull these guys with these other guys'. We want to feed back the matchmaker and that's going to come from this loop that I call, so machine learning and AI are on top of our mind. It's part of the questions we ask when we do interviews these days, and hopefully in the next 12 months, we can get to that point.
And you mentioned hiring new employees for Edgegap. Do you have a ballpark of how many you'd like to have, you know, in the next year or so?
Yeah, by the end of 2020 if everything goes to plan we want to be between eight to 10. It's a mix of junior we have, I need one more executive. I need someone in the sales strategy role. Someone's going to be in more of a product management, business development role. But besides that executive, I want to hire at least one more senior technical person, a few more junior, and we have interns where we have a partnership with a local college here, and some University. Beyond the R&D that I want to do with the University, we're trying to get a lot of interns because they love games, they love the culture, we nourish the culture here. We try to be, I don't want to sound cliché, but it's like a family, kind of. We have a big living room within the studio where we play video games. And we want to work around that and it seems to be working for now. Since we're on the south shore of Montréal, we can attract them with different things rather than 'Hey, why don't you just take the train one hour in the morning one hour in the evening just to go back and forth to work?' No, no, you can just take your bike or your car, five minutes and you're at the office. It's more of a place to live than anything. So we work around that, try to get that going. That's why we have fairly good success with younger talents. This new generation is more careless about just benefits, benefits, and the like. It's more about 'Okay, am I going to have fun or not? Because I have the choice'. There's a lot of jobs out there. That's what we are betting on these days. In fact, a local newspaper is going to do an article about it in the next few weeks.
Cool. And I'm sure you probably have heard of Network Next, which is a company that is kind of trying to be in the same space as Edgegap because they also want to reduce lag. Although they seem to be taking a different approach. I think because they want to us Like reroute traffic, something like that, as far as it goes, understand, so can you help me understand the differences between what you're doing and what they are doing.
Sure, sure. We're both trying to tackle the same problem, as you've mentioned. The difference is that, let's say you have to drive from your house to your workplace and it takes one hour, their approach is, we're going to build a new highway, or we're going to have more lanes so you can drive faster from your house to your office. We're going to take your office and we're going to bring it closer to your house. It's quite a different approach. We do not ask studios to change their games. With Edgegap we work with existing games, there's no need to change anything in the game. There's no need to integrate the game itself with any API or SDK, we only integrate from a backend perspective. Whereas the way I understood it is with Network Next you need to use their code within your game so you can use their peering network. I believe that it looks similar to the Steam peering network, it's like a private network that goes over the Internet that gives you Points of Presence. I mean, most of the cloud already have that. So I think that's what they're doing, but they're doing it from a network perspective. I don't know if they're successful or not. They raised a lot of money. Last time I checked It's not always a blessing, but the founder comes from other studios in the gaming industry, so I'm sure he knows what he's doing.
Indeed. He worked at Respawn Entertainment on Titanfall, for example. I've interviewed him a few months ago and he told me they're also looking at things like Low Earth Orbit satellites (SpaceX's Starlink) to reduce latency in the future. What do you think about that, will it be feasible for gaming?
Yes, I think it will be. I've looked into it and there's definitely something, SpaceX is onto something. The numbers that I've seen were pointing toward very, very low latency between their satellites and whatever centralized data center that they could host themselves. So there is something there. I don't know how long it's going to take though. I was reading an article about those constellations of satellites that were obscuring the sky for astronomers. It takes a lot of those low orbit satellites, it takes a lot of them. Still, this is promising and there is something there.
Anything that can reduce lag for us is good!
Yeah, I agree with that. There are a lot of technical challenges from a technological perspective when you deploy a lot of them and I think I was reading they need, don't quote me here, but I think they need 60,000 satellites to launch a service. They're small and everything but it's a lot of stuff. Still, the numbers that I've seen, if they can hold on to that, man it's gonna be great. The Internet is going to go away. It's going to be purely satellite.
Let's hope it happens sooner than later then. Anyway, with Edgegap, do you work with ISPs or CDNs?
When I initially started, we were planning on using service providers like Vodafone, Telefonica, Verizon. These guys are too slow right now because we are a startup and we need to sell something, get something off the shelf. So we're still pursuing service providers, but six months ago we started to talk with CDN vendors. They already own a footprint worldwide and most of them are starting to convert their infrastructure into an edge computing infrastructure. They're allowing people to use their stuff beyond what I call the one-way communication, right? Streaming is a one-way communication, for example. When you're receiving video or downloading The Witcher 3, you get something from there, but you don't send data. StackPath and CDNetworks are two large CDN vendors and they're both converting their stuff into an infrastructure that's spread across the territory.
One thing that's new as of a few weeks ago is the announcement of Wavelength by AWS. It's like a bit of an outpost, it's an edge computing infrastructure that they do in partnership with service providers, they've announced Verizon and Vodafone. The way I understood it is Amazon is going to ship hardware to these guys and they will just physically put them in their network. And that's going to be edge computing from Amazon.
That's interesting because as far as I know many game studios already use Amazon Web Services.
Yes, yes, but even if they use AWS, the way it's been used today, if you use the AWS game tech, you have standby instances. Take Rainbow Six Siege, and I'm not saying it is hosted on Amazon, I don't exactly know. At some point during the day, you have 100,000 players or even a million players. Let's say you need 50,000 instances, you just need a little bit more to make sure that the people would not have to wait for VM to boot. There's this big notion of standby instances and handling over or managing how many instances at a given hour in a given day you need, and Amazon is trying to bring machine learning into that and say 'All right, we're going to take care of that. So if you need 100 VM, we're going to start 110 VM and try to just be ahead of the curve'.
But you still pay for standby instances, it's not on-demand. It's pre-emptive and that works if you have a limited amount of regions. So if you're Electronic Arts and you need to host game servers in, I don't know, eight regions within AWS, you need 100 VM there 100 VM elsewhere. You do it like that. It kind of scales, right? Because you only have eight regions. What if in our example you don't have eight regions, you have 200 regions, you have 10,000 regions? Are you really going to have standby instances in each of those 10,000 regions? I don't think so, that's not scaling at that point in time. So that's what we're trying to work with. And we believe we're in a pretty good spot at this stage right now.
Something else that's also going on right now is the arrival of game streaming with Google Stadia, Microsoft's Project XCloud, not to mention rumors of an Amazon game streaming service in the works. And I'm wondering if this is a bit of a counter to what you're doing because you are trying to reduce latency while the streaming services inherently add latency, right?
Yeah, I did a post on our Edgegap blog, which is one of the most visited web pages on our website listing 30 different cloud gaming vendors. And beyond the ones that you've mentioned, there are others like Rainway, Shadow. There are also some which are not officially announced. I've heard through the grapevine that Disney may have one, Walmart may have one. I don't know if it's true or not. We did see some info on the interweb. I do believe after seeing Stadia's launch that most of the games are single player because they know they're injecting lag, right? The fact that something else is going to render the 3D, they are injecting lag. So they already deal with that, they're pretty busy with their single player problem. If you add the multiplayer problem, you're having lag. Cloud gaming makes lag worse by three to four times and the reason is, if you and I are playing together, you're going to use this data center that is closest to you and I'm going to use one close to me here in Montreal. We'll have then have to connect to a server to play together because that's how it works, most games work like that. And let's not talk about P2P but even when we're using P2P we would have this problem. So you now have the lag from rendering and the lag from the multiplayer, that's four places. One argument that I've heard from one of the Google engineers was that you're going to be in the same data center.
Yes, I've heard that too.
That's right. If you're from within the same data center, there's probably very little lag multiplayer wise, right? But will your game have enough players at every single point in time in each of their Stadia regions to play the game? So if we take PUBG, are you going to have 100 players connected to Montreal's datacenter that are trying to start a game at the same point in time? Because even if I was playing that game five minutes later, I'm not going to be in your game. The game has already started. So I think it's a weak argument because there are not enough players in any given game. But I know they have a big backbone, they have a big network - after all Google is quite big, right? So they can backhaul the traffic quickly. But I don't think that they figured out how to bend the speed of light yet. So on the cloud gaming front, we're getting attention from these guys in the sense where I have communications with some of them about Edgegap.
And would it be possible for your Edgegap technology to kind of limit some of the downsides of cloud gaming?
Yes, that's what we envision. We're not going to eliminate lag for everyone, you cannot put lag at zero, that's impossible. You can mitigate lag, you can bring everybody below a certain threshold, right? You can say 'Alright, you know what, at 80 milliseconds, it becomes a problem'. So what we're able to do is to put most of the people below a certain threshold, let's say 50 milliseconds. So if you have 80% of your people below 50 milliseconds, they are now having a better experience.
Yeah. Personally, one of my pet peeves with online gaming is that, as you said, you can play with a friend and have decent latency as long as you are in the same country, maybe in the same continent, but you cannot really play properly if you are in two different continents. Can Edgegap help make this happen?
You're right. Two benefits we found of our Edgegap solution after testing is that we can remove the barrier for matchmaking. Let's say you have a very popular game and your game has a lot of expert players, beginner players. And when you do your matchmaking, you don't want an expert player to go against a beginner for obvious reasons. But you break down regions. Let's take North America just for the sake of this discussion. You break down the region into four quadrants and you have a matchmaker per region and you're like 'Okay, if you're in the NorthWest and you need to find an expert with you on the northwest because otherwise, the lag is going to be too big. And if you're in the SouthEast, we need to find an expert to play with you because if it is somewhere else, you're going to be a lot, you're going to get a lot of lag'. The problem with that is that you may end up waiting for 10 minutes before we get another expert to play against you. Whereas with our technology we remove boundaries for matchmaking. So we can look at North America as a whole.
And I can say, all right, I have an expert on the west coast, I have an expert on the east coast. And if we look at that, they can play together because if we put the server in this specific spot, they're going to get 50 milliseconds each and it is going to be great for everyone. So to your example, if you're on the other side of the Atlantic, and I'm here, maybe there's somewhere in between that would make a lot of sense and we would have a fairly decent player experience.
I'm not gonna say that it is going to be perfect as if we were in the same city. But maybe we're going to make it equivalent. So one of the things we're looking at with eSports is to make it equivalent for everyone. So you're going to have the same amount of lag as I am instead of me having 20 milliseconds and you having 200. This is not fair. We want to make it fair. We're working with the Montréal eSports Academy around that concept to make it fair for everyone so we can guarantee that during a tournament or for rated games, everybody gets the same kind of lag.
That's interesting. Certainly, it's one of the biggest issues as you said because as long as everyone in the game has 50 to 60 milliseconds of lag, it's decent and fair. However, is someone has 20 milliseconds of lag and the opponent maybe five times as much, then not so much.
One thing we're also looking at is that we're looking at kicking people off of a game. So we integrate with the matchmaker. And if we see that someone looks to be within the US, but really is using a VPN from Asia trying to kind of connect to a region they're not supposed to, we can tell them the matchmaker 'Hey, don't pull this guy and just kick him so we're trying to interact and close the loop pretty much the same way as we do with machine learning.
Okay, is it hard spotting people using VPNs?
By using actual real-time telemetry, Edgegap is pretty good at that. It just needs more integration with the matchmaker component as most studios have proprietary software.
Last question. Do you have any plans to do additional rounds of funding for now?
We all hope to be profitable at some point. And that's the goal. Considering we're after larger studios at this stage, I do believe we may need another round based on how things are going, but really, I'm coming from a business family and I believe in profitability. But I don't think it's definitely not other loop and we already have some very, very preliminary discussions around that is going to be based on the goals we've achieved in the next six to 12 months. We'll see how it goes.
Good luck, and thank you for your time.