When we blogged about the launch of Apex Legends back in April, the free-to-play battle royale from Respawn Entertainment had just hit the magical 50 million player mark the month prior.
As the quintessential example of the kind of scale that Multiplay’s Hybrid Cloud can achieve, hitting over 2 million concurrent players (CCU), we had to get the Respawn team on camera to talk about the launch and management of the game.
Thankfully, the team were kind enough to let us distract them for a few hours with a film crew and boom mic. You can watch what we came up with below, and read on to learn more about the tech of one of the fastest growing games of all time.
Thanks to all those involved in the video!
- Jon Shiring, lead programmer, Respawn
- Drew McCoy, exec producer, Respawn
- Fred Gill, head of technology, Respawn
- Steven Croop, Multiplay’s customer engineer, Google Cloud
- Paul Manuel, managing director, Multiplay
For a more detailed look at the Apex Legends story, you can download the free PDF case study here.
Planning for the unexpected
As Jon and Fred noted in our conversation, forecasting how many players will turn up to play a free-to-play battle royale is almost impossible. Sure battle royales are popular, and fans love the Titanfall franchise, which Respawn had launched previous to Apex Legends, but how do you ensure you’ll have the server capacity for an unknown number of players?
In short, you can’t. Instead you need to ensure your back end tech is up to the task of scaling to millions. Respawn asked us to arrange for enough bare metal capacity for around 200k CCU, but wanted to ensure we could scale much higher.
To keep up Multiplay’s side of the bargain we ran a scaling test, which we blogged about at the end of 2018 (without naming the then secret Apex Legends of course!).
In this test we essentially requested 120,000 cores from Google Cloud, enough capacity for 14M players, in the space of a couple of hours. Multiplay’s scaler stood up to the challenge, giving the Respawn team confidence they could launch and hit big numbers without a hitch.
At peak, the number of players was well over 2 million, 10x more than Respawn’s initial forecast, so the testing was worth it!
Many games launch without ever knowing their full potential because they run out of server capacity and people are put off trying the game out, perhaps forever. Thankfully, because Apex Legends was able to benefit from Multiplay’s integration with various cloud providers, the player count kept on climbing.
Post-launch management of Apex Legends
As you’ll know if you work in gaming, the work doesn’t stop after a game has launched. This is certainly true for the Multiplay team which, in addition to managing the game 24/7 globally, optimizes Respawn’s blend of bare metal and cloud.
The focus here is on maximizing cost-efficiency, without affecting server performance. Once player numbers become more stable, after the initial flurry of players, the Multiplay team is able to work with Respawn to put in place more cost-effective bare metal servers.
While cost-effective, these servers are still highly performant. Sourced from Multiplay’s trusted network of data centers, each and every one is QA’d before use.
When Respawn rolls out new content, such a new map or mode (we loved what they did for halloween), and there is a spike in player numbers, the scaler once more bursts into the cloud to ensure there is enough capacity.
This means players get a smooth experience and Respawn are able to manage their costs.
Diving into the tech: Respawn’s back end infrastructure
As you can see from the diagram above, the back end infrastructure of Multiplay is heavily integrated with Respawn’s. This has evolved over the years, but the fundamental relationship between Respawn’s matchmaker and Multiplay’s servers, via the API, remains the same. We explore this more in the case study.
Want to learn more?
To learn more about the back end infrastructure, and for more insight into the launch and management of Apex Legends download the free PDF case study here.