We are excited to bring Transform 2022 back in-person July 19 and virtually July 20 – 28. Join AI and data leaders for insightful talks and exciting networking opportunities. Register today!
If anything will stop the metaverse from being an enjoyable experience, it’s latency. Latency is a measure of the delay that happens during interactions between two parties over things like the internet. And Syntropy is trying to attack this problem so that the real-time internet, and eventually the metaverse, can happen.
Syntropy recently teamed up with sports betting and gaming company Entain to launch a global routing protocol based on its blockchain-powered relay network. The aim is to reduce latency for metaverse and web 3 applications through decentralization. The protocol is designed to detect performance problems and switch traffic to alternate paths.
Syntropy has a decentralized autonomous routing protocol (DARP), connecting datacenters across the globe. Syntropy’s DARP technology runs on top of the public internet and is fully programmable. The global digital fabric creates a network of networks composed of the best paths, said Domas Povilauskas, CEO at Syntropy, in an interview with GamesBeat.
Following a technical test, Entain has begun using Syntropy across its 40-country customer base to deliver better performance for users. Entain’s tests found 1,500 milliseconds of delay for 35 minutes over its default internet path. DARP technology detected the bottleneck and routed traffic around it. I spoke with both Povilauskas and Syntropy’s chief revenue officer Jeffrey Lacouture.
Here’s an edited transcript of our interview.
GamesBeat: How did you start the company? What are you trying to accomplish?
Domas Povilauskas: We started this company in 2018. We’re almost going on four years now. Right now we have 56 employees. The company is fully decentralized. We don’t have an office. The business development team is in New York. Marketing is in the bay area. I live in Los Angeles. Then we have the engineering team spread out over in Lithuania, where I’m originally from. That’s where a lot of talented network engineers are being developed.
Since the beginning, we’ve been building a distributed, autonomous routing protocol. If you’re aware of BGP, that’s the protocol that powers the public internet routing system. We’ve created a protocol based on a distributed ledger technology that connects the nodes, the computers, and the servers. It’s a single mesh network. Then those nodes share their performance information between themselves, forming a global routing ledger. This global routing ledger always knows the best path – the lowest latency, the lowest packet loss, the least jitter – between any two endpoints in the mesh.
What it means–right now we operate a network of 1,000 nodes. These nodes are located in virtual machines or parallel servers. They’re being hosted in 159 cloud networks. When we connect these nodes into a single mesh, it means they can start sharing information with each other, and they form a global routing ledger. This routing ledger gives you a deep performance insight into every cloud network that is connected to the mesh.
Once you have that, you always–the ledger we have right now, you can see this live. This is the mesh. These green points mean that there’s an optimization. There’s a better path than the public internet if you route through a relay point in the mesh network. I just took a random point in the mesh here between Melbourne and Zurich, from the Oracle cloud to a private data center in Zurich. Public internet latency is 270, 280, and 290 milliseconds. Our latency is 235, 245 milliseconds. We reduce latency by around 40 milliseconds. In this route here, where the public internet latency is 300 milliseconds, that’s basically non-productive. We can lower it to 240 and make that usable. We enable this link to operate while the public internet case, no application could run at 300 milliseconds of latency.
There are optimizations where we can bring it down from 150 to 90 milliseconds. There are various different use cases for how we improve. I’m just choosing random points here.
Also read: Bringing co-creation to the metaverse with experience design
GamesBeat: What you described sounds familiar, based on interviews I’ve done with other companies. I’ve talked to Subspace a lot about what they do, and Network Next as well. Even WTFast. Subspace was trying to both locate dark fiber and also route around known bottlenecks using their own hardware, a lot of their own investments.
Network Next didn’t own their own hardware, I believe, but they did try to map the fastest routes and then figure out pricing in between each of those routes in some way that could enable gamers to get faster internet connections. One of the cable companies in the U.S. was also introducing a lower latency service.
What’s unique about this approach? What does the decentralized approach get you?
Povilauskas: I know Subspace very well. I know the co-founders personally. The main difference between a centralized model and a decentralized model, a web 3 model, is that we create an open connectivity system. Anyone can connect their machine to the mesh and become a relay point for others relaying traffic. That’s powerful because we’re creating a network of networks. We’re not creating another private network.
That’s why Subspace invested tens of millions of dollars into buying equipment and everything, but they’re trying to compete with all these other networks. Google is investing. Amazon is investing. All these billions of dollars are going into network resources. Our approach is totally different. We’re connecting all the world’s best networks into a single mesh, and then providing the user a single interface where they have access to this network of networks, almost like the public internet right now. But they have the ability to route traffic along the best possible path given the price you’re willing to pay for your traffic.
The core idea is about what we call an open internet exchange. Any node in the network is able to connect to some piece of the infrastructure. Currently, we’re talking about institutional nodes like data centers, cloud providers, and ISPs. We’re working with some of the large ISPs in Asia. They’re providing us with nodes in the network. This is not public, to be sure. But we’ve connected 159 cloud networks. Now we’re connecting more underlying networks, including ISPs. We’re forming this mesh of routes. We do not own the infrastructure. We either work with partners, or anyone can run a node. If you want to host something in a data center and you think it’s a good relay point, or you’re buying straight from the ISP, you can put that on our exchange and sell your internet connection to someone that wants to use it as a relay.
This is a unique approach. While we sell to customers, we never–most of these enterprises we work with already have some network solutions they’ve implemented, like SD-WAN and other recent technologies. But what they don’t have, they usually always stick to one vendor. You have to stick to one vendor. You have to buy from the ISP. You’re always dependent on that relationship, in a way.
GamesBeat: What’s the result?
Povilauskas: What we’re saying, we’re offering this network of networks. We connect Google, Amazon, all the world’s networks at once, and we ride other backbones. You just get a price for that. With this use case we have with Entain, we just announced in this case study–they avoided a 40-minute outage. They had a good network, a dedicated network that they managed. But there was an outage in the transit. Their route was gone. The latency was something like 1,500 milliseconds. Their applications stopped. Some of our routes experienced the same outage, because it happened somewhere significant. But some of our routes were fine, so we just rerouted the traffic.
We always have at least five alternative routes. We’re always able to use different networks, different backbones to route traffic. This is how we see ourselves, as a global connectivity fabric. Not a new private network that’s investing and competing with Amazon and Cloudflare, but a network that connects all these networks, provides a very friendly user interface, and you can connect any nodes on the internet instantly to have MPLS-quality connectivity that’s always dynamic, always choosing the best route.
GamesBeat: The bottlenecks are where?
Povilauskas: You don’t have to deal with ISPs. This is the big thing. We’ve had conversations with Meta, Google, all these major companies, and the main challenge they’re dealing with is their relationships with ISPs. ISPs own the infrastructure, but the ISPs are motivated to sell access and get profit. Content providers are motivated to sell content. Content providers are motivated to have the best possible connectivity for their end users. Meta right now is trying to build their metaverse. They’ve publicly said that one of the largest challenges in building the metaverse is connectivity. In order to make it a reality, there should be drastic changes in the telecom system. The metaverse requires super-stable performance, and content providers don’t control the networks. That’s why they’re trying to invest a lot of money and build these private backbones. But they still rely on the ISPs.
We’re creating global connectivity coverage, this network of networks. This protocol sits as a mesh that you form. We override the ISP routing system. I don’t need to work with ISPs anymore to get global access to all the world’s networks. You just connect to the mesh. The mesh is based on an exchange where every node can sell and relay traffic for others. You send traffic through a relay. When we introduce a relay point–the last example I’ll give, say you’re on the Verizon network and you’re sending information to Europe. You’re accessing information from a European network like Deutsche Telekom. The only way you can access that DT network is the way Verizon chooses to reach it. You don’t have control over the way Verizon routes.
What this network allows you, if there’s a relay point in Sprint’s network or AT&T’s network or any other network, we can force traffic to that network and send it on to your destination. You bypass the ISP routing system. You can jump in between the networks as you want, depending on the performance they give you. This is why it’s very powerful.
GamesBeat: What stage are you at right now? How far along is the company to where you want to go?
Jeffrey Lacouture: Our product hit general availability – the product Dom was just describing – last year. Entain is the first customer we’re announcing.
GamesBeat: What kind of service did they need?
Lacouture: Entain is a European gambling giant. They’re an FTSE 100 company. They run about 20 different brands, everything from Ladbrokes to Bwin. They have 25,000 employees. They run a joint venture with MGM called BetMGM that you may have heard of. They have a heavy European presence and a big American presence as well.
What they use Syntropy for is specifically latency improvement and outage avoidance between their data centers. They ran a proof of concept with us where they installed our software in two different data centers, and they found that we could help them avoid outages when there was a downstream provider that had some kind of congestion, or as Dom described, some kind of provider to provider relationship problem.
GamesBeat: When I was talking to Comcast, I got the sense that if you went from a network of one gigabit per second to 10 gigabits per second, it wouldn’t really impact latency. If you want to attack that, it’s more like–one thing they were doing was classification of traffic. If you have a game that’s real time you prioritize that. If you have video you prioritize a continuous connection. Everything else waits in line.
Lacouture: And that’s like what WTFast does as well. They’re able to identify where the critical moments in the games are.
GamesBeat: I guess the question then is, what can these companies do that is the main thing to attack latency?
Lacouture: This is the thing. Latency is a tough thing to stamp out. Entain is global. They have data centers in the U.S. and in Europe. They have communication between them all the time. It’s possible that you’d have a user in California logging onto an application hosted in California – because a lot of regulations require that online gambling has to be hosted in the state – and then the backend, the thing that supports that, maybe the authentication for the user or an account activation, that happens in a data center in Europe.
Now the packet has to go from California to Europe. It doesn’t do that just once. It might do it 10 or 15 times. You have transactions going back and forth. When you sit there on your phone and push a button, you don’t know how many transactions are happening underneath the hood. The thing that Entain has said publicly is that they process 71,000 different bets every minute. I’m not a big sports bettor, so for me, it’s just putting $20 on the team I want to win before the game. But serious sports betting nowadays is betting on every play, every at-bat. You can do all that in real time. Every minute they’re updating all their lines for all the bets you can possibly make, and all these transactions are happening on a global level.
What’s really happening here is that latency, when it spikes to more than 100 or 200 milliseconds, it starts to impact the speed with which they can provide those bets. You can think of it as the stock market. But it’s 24 hours a day, seven days a week, running globally. That’s the problem they solve. They take Syntropy, put it in the path, find lower latency links, and improve performance for their network.
GamesBeat: I’ve talked to people on the game side about some of these challenges. The Fortnite people, like Kim Libreri, the CTO at Epic, brought up the sniper problem. You can solve that problem with how you design your game maps. But if you have a sniper on top of a mountain, he’d normally be able to see into a bunch of grids that are normally on separate shards of a large game. You can have low latency within a shard quite easily, because the players interacting with each other in that space are together in one place. But if you have a map so large that there are multiple shards, the connection between the sniper and his targets then becomes a problem. You don’t know which person the sniper could go after among all those different players.
And they’ll just say, “Right, that means we can’t build a map that big.” You put in trees and mountains and other things that obscure the view so no one can see that far. But when we start to think more about the metaverse and more people in the same space, that’s something they would eventually want to get to. Does that problem get solved at some point? Is it just a matter of putting a lot of CAPEX into the internet? Or is it possible today?
Lacouture: I’ll answer the question directly, and then I’ll take you a little bit past where we are today toward what we’re thinking about. The answer is that lower latency is better, always. It’s better for certain applications, more extreme applications. Gaming is one of them. Stocks, like I talked about, that’s another one. Gambling, absolutely another one. There are all kinds of places where that happens.
Then there are places where latency maybe doesn’t matter as much. Content delivery. Latency is not that important for content delivery. If my Netflix is 500 milliseconds behind yours and we’re watching a live event, it doesn’t matter that much, especially since most of the content is time-shifted. There are applications where it matters. Gaming, Fortnite, sniping, are absolutely a focus. When it comes to metaverse gaming, because of the way it works, latency is incredibly important.
The thing that Syntropy does is it doesn’t matter where you are in the world. It doesn’t matter how many different servers you’re potentially trying to connect. Syntropy can provide you with the fastest path mathematically possible across the network. Right now you cannot do that. Right now, you’re beholden to the provider to give you a low-latency link. Maybe you have an agreement with a provider to give you that low-latency link and they provide an amazing connection. Or maybe someone is coming in on their home connection. Or maybe you have servers that are far apart and there’s an outage in between.
What that comes down to is that you are not in control. Whether that’s you as an individual user, you as an app developer, or you as someone running a data center, you are not in control of how your internet traffic flows. When you can use Syntropy, you can say, “This is how I want to control my traffic. I want it to go through this node.” All of that is automated by our system, but if you really wanted to, you could say, “I always want traffic from my Los Angeles node to go through my Chicago node. I always want to use the Cogent link.” However you want to define it. We are giving people programmable internet that gives them performance. What you see in the gaming application, that can become very powerful, because of the problem you just described.
Povilauskas: Some time ago I was thinking about the metaverse. I’ve read a lot of articles online. There was a good Gartner report talking about “What is the metaverse?” Everyone’s hyped up. The metaverse is very similar to gaming. It’s almost the same from an infrastructure perspective. People imagine that there will be one virtual world, a digital twin of this world, where we all connect and somehow interact and have virtual experiences, a virtual life. But if you look into the technical side, the infrastructure, how the metaverse should be built, you understand that the metaverse will not just be one world. It’s going to be a world of worlds, a million worlds connected together.
Latency depends on physical laws like the speed of light. If you’re in the United States, your friend is in Europe, and you want to have a metaverse experience with them in VR, that’s going to be hard. There will be physical limits. You can’t just teleport the data. Right now it’s just the physical systems we have. There will still be latency.
How the metaverse will actually happen is that thousands of companies will create these worlds. There will be the Facebook America world, Facebook Europe, Facebook Asia, Facebook whatever. There will be different Google worlds. Different worlds launched by thousands of companies. They’ll be connected almost like gaming servers. You’ll change from world to world, and you as the user, you’ll just have your identity. Your avatar, your features, your digital objects, your digital property will be on the blockchain and you will be able to transfer that digital identity to any world you go into. If gaming right now is big, the metaverse is going to be a thousand times bigger. You’ll have a thousand times more worlds, more servers, more people connecting together to play games, to chat, to have coffee together in a virtual world, to shop.
Right now Facebook’s saying that their largest challenge is the network. There need to be drastic improvements in the network. You can see a bunch of articles from Meta that outline their problems. And they have the capital, billions of dollars to invest. But if we’re talking about this world of worlds, where every company will host something, every shop presenting a virtual shop experience–almost every company that wants to host something in the metaverse will become a gaming company, in a way.
Where is this infrastructure going to come from? Facebook, sure. Facebook has billions of dollars to do that. But it’s still challenging. We believe that we are creating this unified global infrastructure so that no matter how big you are, whether you’re Facebook or a small company that wants to launch its virtual world into the metaverse, you can just connect to this global internet resource. We provide access to the global internet network, the same as you would if you invested billions into building the world’s largest backbone. We provide you the same access as Facebook has to the whole internet’s telecom resources. Then you can host whatever you want to host.
In the end, Syntropy is not only enabling lower latency. The latency problem will eventually be solved by 5G, or improved at least. Again, we have physical limits, physical laws. There will be physical solutions to latency. We’re creating a different model for telecom in general. This is about access to the internet, access to the network. What access do you have right now as a gaming company, or if you want to launch your metaverse? If you don’t have billions to invest in your infrastructure, what kind of access do you have to the internet today? You have very little. We provide that access. We’re a system that opens up the internet infrastructure for everyone.
Regarding our customers, as Jeff said, Entain is definitely one of our largest, but we have more. We have more enterprises. We have around 2,500 accounts right now that are using Syntropy in various ways, from very small accounts up to larger accounts that send much more data and use the Syntropy network more profoundly. In total we have around 3 million connections over the system already. I need to check the deck for that. That’s 3 million routes to route data. It’s not as if we just have one customer. Entain is only the first customer we’ve announced publicly.
GamesBeat: What are you hearing from the gaming companies?
Povilauskas: We’re hearing very positive feedback. We have very strong demand. We just went to GDC a week ago and had many conversations with the game industry. We have many leads coming in, even companies like Cockroach Labs. You mentioned databases. Integrating with CockroachDB, where they provide database infrastructure for gaming, together as a solution we can provide a full scope of performance.
The application of this technology is pretty broad. We can sell the same technology to various different companies. That’s why we established an enterprise team. Jeff is leading that enterprise sales team. We have long-time sales professionals from that space. We chose the enterprise angle at the beginning of our marketing. Right now we have Entain and other enterprises as yet to be announced. We chose enterprise as the go-to-market, the mission-critical applications that need it. But gaming right now is becoming the second one, the second big angle that we’re entering just now.
GamesBeat: And then there’s the metaverse waiting.
Povilauskas: The metaverse is pretty early right now. But my personal belief is that in the next year or two from now, the next big bull market is going to be about the metaverse. It’s going to be everything. Everything we have here is going to have a digital twin in that world. The question is, what does the metaverse need to become reality? People are hyped up and talking about it everywhere, but let’s debunk it a little. What does it actually take to build this?
GamesBeat: I don’t know if anyone has done a good assessment, but I remember Intel’s chief architect saying that it’s going to be a 1,000 times increase in computing power in order to enable the real time internet.
Lacouture: One thing that keeps coming up in the metaverse and things like that is if you are going to have people spending more time online, more time in virtual environments and things like that, that data doesn’t just live in a data center. It lives across multiple data centers and multiple client-to-data-center connections. Right now we’re in a Google Hangout. How does it work? It goes from my computer to Google to your computer. If we instead put tens of thousands of people in this space, it can’t be centralized. It will have to be more peer-to-peer.
That’s where a technology like this comes into play. I can build an encrypted connection to you directly. We can find the most performant path between us, which is going to be important if we’re on a video call.
Povilauskas: And 5G is peer-to-peer as well by its nature. The metaverse is possible only through that peer-to-peer, in a way. Centralization is always a bottleneck.