Chivalry II: Managing player spikes with zero downtime
How we helped Chivalry 2 manage double the expected player count on launch.
Launch day can make or break a game. Especially for multiplayer titles, where players expect a trouble-free game and don’t have much patience. (Let’s face it, there are many games to play.) So when Torn Banner and Tripwire developed Chivalry 2 – the sequel to their medieval multiplayer hack-and-slash game – they knew they needed scalable multiplayer servers. Ones that could handle a huge influx of new players and make the launch experience as smooth as possible.
So they turned to Gameye to make sure they had the systems to cope with any spikes, on five platforms in 30 locations across the world. Our strategy was to help them ‘scale with ease’ and be prepared for any eventuality. We made sure there wouldn’t be any capacity issues. And that they could add regions with no financial risk, while integrating the rest of their tools.
We handled a surge of players on launch day
Players flocked to Chivalry 2, surpassing our internal predictions. There were twice as many sessions than we anticipated in the first hour, and over 250,000 players at launch. And almost a million players in the first month.
While there were launch issues, they were all code-based and not related to our technology. On our side, our servers had no downtime and no major issues.
The need for a new multiplayer solution
During the early stages of creating Chivalry 2, Torn Banner – the developer – and Tripwire – the publisher – knew they needed new multiplayer servers. The system they’d used for the first Chivalry game left them overpaying for idle machines and unable to add new servers on the fly. In short, they were expensive and couldn’t scale easily. The current servers weren’t going to work for Chivalry 2, so they went in search of a new solution.
Gameye hosts multiplayer servers and is completely platform agnostic. We manage the server capacity by giving developers a single server farm, which can scale quickly, affordably and globally. Developers use a simple API that ties into the game’s matchmaker. Game sessions start when the player clicks, using our proprietary Docker-based orchestration. And when demand surges, we automatically allocate more resources to the game. Developers don’t need to guess how much capacity they will need or worry about inflated cloud hosting costs. We take care of everything, with a predictable resource-based pricing model.
From the player’s perspective, the experience is the same. For the developer, it cuts costs and they can use whatever other technology they want. It also takes the burden away from them, so they can focus on creating new content or dealing with other moving parts.
Innovation takes a bold step
While Torn Banner and Tripwire were excited about using new technology, they were also nervous about whether it would work. It’s always a risk to choose innovation, but they were frustrated with the old way of running their servers.
“When we spoke to Gameye, I’ll be honest, we weren’t sure it could be done,” said Mike Stone, the senior producer at Tripwire. “First and foremost, we were attracted, not only by the fact you were able to considerably lower the costs, but that you were able to scale so quickly.”
It was a big risk. As Chivalry 2 is a multiplayer game, the servers were their lifeblood. They needed to find a partner they could trust to run it all.
"I don't think we would have had such a good relationship with anyone else."
A partnership built on trust
On paper, we offered everything from a technology point of view, but technology alone wasn’t enough.
“What attracted us throughout the process was that you were attentive, easy to work with and you’d be on a call when something came up. That was comforting to us – that you’d be on hand,” Mike said. “The relationship has to be a two-way street. Even if we didn’t agree on a specific issue, you were always willing to see it through. There was always room for conversation.”
We had built a platform that would be perfect for Chivalry 2 – with data centres set up in 30 targeted locations and able to reach 250 data centres in total if they needed. We can automatically scale, allowing their servers to be flexible.
“We felt there was a personal relationship, and if there was a problem, we knew you guys [Gameye] would be there,” said Rasmus Lofstrom, game director at Torn Banner. “I don’t think we would have had such a good relationship with someone else.”
Test early, squash bugs
“We knew we needed to test early, so that we could optimize the systems for the number of players they’d get,” said Elmer Bulthuis, CTO at Gameye. “It took us just a few days to get set up. The important thing when you’re choosing innovation is that you need a guide through the whole process. So we walked Tripwire through how to integrate our systems, step by step, so that they knew it was all working correctly.
“In total, we ran six alpha tests with real players. These were closed tests to make sure that people could join the server, play the game and do so without any downtime. Each time we increased the number of players to make sure the servers could handle the load.”
“We did have some issues with performance to start with,” said Rasmus. “It was inherently frustrating – as any of these problems always are. We only sometimes saw it in the alpha tests, and we initially assumed it was on our end. But as soon as we raised it with Gameye, you guys jumped on it immediately and solved the issue in about two weeks – ready for the beta.”
Dealing with last minute changes
It’s difficult to predict potential players and Tripwire had noticed a huge increase in the response to the game. They re-evaluated their numbers and the new projection was much higher than we originally anticipated.
Success can also be a multiplayer game’s downfall, if not handled correctly. If you can’t scale to meet the demand, it can kill the experience. In two days, we’d made sure the server farm was ready to accept the increase in players.
“It was really appreciated,” said Brian Etheridge, Tripwire’s publishing director. “A lot of times scaling server capacity at the last minute can cause loads of problems. But in a matter of hours, you’d managed to spin up new servers. We were ready for our open beta.”
Launching a successful multiplayer game
On 8 June 2021, Chivalry 2 was released globally, on five platforms. In the first hour, the influx of players was almost 100% more than projected, a huge surge in players to jump in at the start. It had over 250,000 players at launch, which levelled out to nearly one million players over the first month.
“We didn’t expect such a large spike at launch,” said Brian. “We thought we’d see a much larger peak during the open beta than on launch day, as it was free to play. But there was a substantially higher number of players than we anticipated – almost double.”
There were some code-based issues at launch, but in terms of multiplayer servers there was no tracked downtime. Latency was low and thousands of game sessions started and finished smoothly on our server farm.
From a server infrastructure perspective, things had performed as planned. Being able to scale quickly is key to our business. They only needed servers in 30 locations at any one time. We have servers in around 250 locations. This meant they could rapidly fire up new servers and decide exactly where they needed the most capacity.
Our philosophy helped reach this point
We always plan for the best and worst-case scenarios. Ultimately, that meant Torn Banner and Tripwire could rely on our multiplayer server infrastructure, knowing that it could, in principle, handle anything. And that was one less worry for them during launch.
If you’re developing a multiplayer game and want to keep your costs down, while making sure you can adapt to any sudden player surges, then we’d love to speak with you.
Get in touch to learn more
Get in touch to see how you could start saving on your server costs.
Case studies form
"*" indicates required fields