Futureplay Games delights players with data-driven design

INTRODUCTION

Digital gaming company embarks on migration journey

Helsinki-based mobile games studio Futureplay began life in 2015. A start-up conceived by five individuals passionate about creating great games in a low-hierarchical work environment. It has since grown, having been acquired by Plarium in 2021, and now has a team of 73. Among its portfolio of games is the Merge Gardens ‘puzzle game’, combining Match-3 core with merge progression. The game underwent a rebranding in January 2023. Though players had enjoyed the mechanics and feel of Merge Gardens since its launch three years previously, it needed a narrative and play metrics had hit a plateau. With the rebrand came the introduction of a strong story arc, a cast of characters, plot twists and enhanced UI. All supported by a series of impressive CGI marketing videos broadcast on social media.

“And it’s taking off,” says Jarkko Laine, tech lead at Futureplay Games.

THE CHALLENGE

Managing live operations quickly to enable growth

On the back of the rebrand, and impressive teaser videos, installations of Merge Gardens by new players began to rocket. Player numbers are currently around 10 million. What Futureplay lacked however was its own server backend, a means of managing live operations within the game in terms of campaigns, offers, events for players, and bringing those together.

Since the launch of Merge Gardens, the company had relied on a third party software-as-a-service backend for storing player state (player name, score, game history), configuring campaigns, and keeping the game running. This approach was effective before the rebrand, as the small staff allowed Futureplay to deploy updates without the need for a dedicated server team or extensive server management.

“Third party solutions were great when there were few people in the team - it lets you get started quickly,” explains Berkay Uckac, server-side software engineer at Futureplay. “And with the scale we had back in 2020/22, it was fine. But with millions of players and lots of marketing attraction, we want to run tests, we want to update quickly, and if those third-party solutions don't have support out of the box, then you start to hit limits.”

For the studio to iterate faster and fully optimize the clear growth potential of its most popular game in a market that’s all about keeping players engaged - it needed to bring this core tech in-house, so it could build on it in the future. “We also needed to get all the data in, on our own systems, and bring those pieces together so it would be easier and quicker to manage,” explains Jarkko.

“So that's where this journey of ‘let's try to make something’ began,” adds Berkay.

THE SOLUTION

Bringing data in-house for optimum control

Resolved to migrating from their third-party solution, the choice of database came quickly for Futureplay. “MongoDB felt natural for us,” explains Jarkko. “The nature of the data in a game logically fits the document model - you have a player, a player state, and all these pieces in it.” Thanks to previous experience and research, such as extensive load testing, Jarkko knew that MongoDB performs well in terms of scalability and handling big sets of player data. Conversations began with MongoDB in April 2023 and Futureplay released the first version of its server in November.

It utilizes two databases in MongoDB Atlas on Google Cloud. The first is an operations systems database, designed for campaigns and configurations, which stores details such as campaign start and end times, and which players to show it to. The second is a player database, where data like player login information, player sessions, and account information are stored. The next step is to bring in more player data, like the player's in-game progress, from the third-party solution into MongoDB - a transition that will happen in the next year.

If there are challenges, it’s that “we cannot just flip the switch, turn off the game and move things completely,” explains Berkay. “It must be a progressive migration - we don’t want to disturb the players’ flow.”

It has, however, been an easy exercise. “A big benefit with MongoDB Atlas is that we’ve been able to manage it ourselves, with just a little help,” says Jarkko. “Things are doable through the nice UIs and terraform infrastructure as code definitions.”

In preparation for launching the backend, Jarkko and Berkay performed heavy load testing, getting things up and bringing them back down, to observe how they happen and determine the right scaling. “And it's been really easy to do those things,” says Jarkko.

He continues, “A big reason why we wanted to go with MongoDB was that once we start doing the bigger migration, bringing bigger player data over, we can then manage the scaling, manage the sharding and get more instances of the database up with MongoDB, rather than having to hire a team of site reliability engineers to maintain a database for us.”

THE RESULTS

Enhanced player installation and experience

It is of course quite early days for the Futureplay team and their new server, but initial deployment of MongoDB went ‘smoothly’. “We utilized private endpoints from the MongoDB side which let us consume data from MongoDB clusters within our own servers securely and privately,” observes Berkay.

A notable backend to end-user impact is that product managers and designers are building and utilizing the player experience dashboard. Here they’re able to manage the game and design the kind of events players get, offers they have, and adjust the balance of the game. “It's easily doable,” says Berkay. In the future, on the design side, the team looks forward to delivering actual data to the players whereby they can have more engaging and fresh experiences. “That’s one of the benefits of having something in-house,” says Berkay, “that you can build on top of it constantly.”

Monitorability has been a big plus. “With infrastructure as a code,” says Berkay, “I can define the cluster I want without the hassle of managing it – but I can monitor it. I can see what the CPU usage is, what the memory usage is, and what the input/output operations per second (IOPS) are, so I know what kind of performance I'm getting out of my cluster.”

The inevitability of change in the software world has made MongoDB’s adaptability reassuring for the team. “We don't know what the future will bring,” says Jarkko. “When setting up databases, we have an idea of what the right structures could be - but with MongoDB it's quite easy to modify those things if we don’t get it right the first time.”

Commensurate with its growth, Futureplay is looking to hire an additional server-side software engineer with a DevOps mentality, “Because even in a small team, we can code, think about the architecture, think about deployment - do the whole thing. The way we use GCP and Kubertnes and MongoDB Atlas - it gives developers access to things previously done by site reliability engineers or specific professionals.”

“For us as developers – our happiness with MongoDB has been key,” concludes Jarkko. “Being able to run this set up, hands on, with a small team – it’s been great.”