There’s a good reason why studios develop multiplayer games – it gives them an opportunity to work at their most creative, building live communities and taking on the challenge of real-time action. They also have the greatest potential to deliver profit. But it’s well understood that they are hard to develop, operate, and maintain thanks to constant updates and trigger points. As Manuel Karg, CEO & Founder of zeuz sees it, the benefit of premium single player games is that they’re built to be client-ready – they can be downloaded or installed onto a device or platform. “You don’t have to have all these player services, you don’t need to matchmake yourself with yourself.”
Multiplayer games are more dynamic. They need a higher degree of human interaction, and in many cases, they also make use of server-side authorisation. When a player does something, it needs to be verified by a server before being communicated to the other players in real time. And then there’s the issue of latency: in fast-paced multiplayer games, the latency tolerance must be below 50 milliseconds.
It’s unavoidable to talk about multiplayer games without raising the issue of cost. With a premium game, studios know that when they launch they’ll get a certain percentage of sales upfront, which will dwindle over time. There isn’t that degree of certainty with a multiplayer game. Its success hinges on two factors that are hard to calculate – the operating costs of the game and its predicted popularity. “Studios try to address this risk from a couple of angles,” says Manuel. “They may be extremely passionate and try to learn everything on the fly. Or they might bring in industry experts. These are people who have experience setting up complex environments and can prove that they have mastered that. The trouble is that there are very few people in the world that can claim to be experts.”
The trouble is that there are very few people in the world that can claim to be experts.
In a bid to avoid these issues, some studios have gone down the peer-to-peer route. Rather than using dedicated servers to run multiplayer games, they use mobile technology to launch games and make matches. But often these kinds of games don’t handle high player counts of demand very well and are only effective for four players at most. They also aren't all that fun for the user, because they are not very secure and are more open to hacking and cheating.
So how can studios get a better handle on the costs associated with multiplayer games? There are two principal ways. One is to model future scenarios and look at the level of consumption. In making these calculations, studios should bear in mind that every game has its own unique design features. These features can have a big impact on the amount of CPU, RAM and bandwidth consumed during a game. If the level of consumption is found to be too high in different scenarios, the studio can then change the game server build to bring down costs.
While relying on the cloud might eat up your budget, relying on bare metal alone might not allow a studio to scale up fast enough if a game becomes popular.
The other key to not breaking the bank is getting infrastructure right. You can either host a multiplayer game using a bare metal server in a data centre or using the public cloud. Going down the bare metal route involves a substantial amount of investment – studios must commit to renting a machine for at least 30 days. But with the public cloud, there is more flexibility; studios can rent the capacity they need per hour or minute.
However, ‘cloud bill shock’ is common. Depending on your server type, bare metal can be over 50% cheaper than renting capacity in the cloud, mostly due to the cost of bandwidth for many players on one machine. For example, if you have 200 players per machine, with a game using on average 150 kbit/s, 70% of your cost will be from bandwidth.
While relying on the cloud might eat up your budget, relying on bare metal alone might not allow a studio to scale up fast enough if a game becomes popular. As the launch of a game is rarely predictable, using hybrid cloud offers the best way of keeping costs down. “If you combine those two things, you can put the estimated base load, average peak load of players, and average consumption onto bare metal and scale the remaining players seamlessly into the cloud,” says Karg. What’s more, post launch, hybrid cloud allows you to maintain an ideal blend of bare metal and cloud – keeping your costs under control.
As you can see, calculating the costs associated with multiplayer games isn’t easy. It’s one of the reasons RETO MOTO partnered with us to launch its multiplayer game, Heroes & Generals on Epic Games Store. After consulting with Improbable, RETO MOTO decided to dispense with its seven providers and move to a more scalable solution, migrating to IMS Game Server Hosting and integrating with our hybrid cloud orchestration technology. The studio is now able to access all the compute it needs from a single provider and can bring together the cost-efficiency of bare metal with the flexibility of cloud. More importantly, it now has the confidence to face any potential scenario.
Finding the optimal mix of bare metal and cloud is no easy task – and should be tailored to exactly what you need. Improbable’s constant monitoring of live data allows us to predict the cost of your project and monitor infrastructure performance, and together we can analyse your current usage and cost structure and compare it to our internal benchmarks. It’s a proactive approach to give you the best possible infrastructure cost for your game.
Keep in touch with our team if you’d like to find out more about effective cost management in game server hosting.