
5 years building a game backend-as-a-service: What we've learned
Chris Wilson •
Making games means crafting experiences that captivate players. But behind every top-grossing title, hidden beneath the art, design, and mechanics, lies a complex technical foundation that determines whether a game can truly scale. This is the world Teemu Haila, Co-Founder and CPO of Metaplay, has been immersed in for the past five years. His mission? To solve one of the industry's most pressing challenges: building a game backend that can withstand the pressures of success.
This is a recap of a longer email Teemu sent out directly to Metaplay Tech Talks subscribers. To get the full version direct to your inbox, with new insights every two weeks, sign up to Metaplay Tech Talks here.
“Building a backend is one of the hardest technical and strategic challenges in game development,” Teemu kicks us off with. “If I’d known five years ago what I was about to get myself into, I might have picked an easier way to spend my working days.”
The Real Problem: Scaling Pain
When Metaplay started, backend solutions already existed to help developers launch online games. But launching was never the problem - scaling was. “The moment a game climbed the top-grossing charts, teams would hit a wall,” Teemu explains. “The backend that got them there would suddenly start buckling under the weight of reality.”
The industry had seen it happen time and again: a studio would find success, only to face two painful choices - rebuild their backend from scratch or patch together temporary fixes and live with growing inefficiencies. Neither was an attractive option.
What struck Teemu and his team was that every successful game studio - from Supercell to Small Giant - had independently built their own backend solutions, and despite using different tools and technologies, they all arrived at fundamentally similar architectures. “It became clear that there was an optimal way to build a game backend - one that allowed flexibility, scalability, and full control from day one.”
Lesson #1: Programmability is Non-Negotiable
In the past five years, few lessons have been as resounding as this one: a backend that developers can’t see, modify, or extend is a backend that will eventually become a liability. “If your backend forces you to work around rigid APIs or wait for a vendor to support a critical feature, you’re already losing,” says Teemu.
Live games are living, breathing entities. They evolve, adapt, and grow in ways no developer can fully anticipate. The only way to ensure a game remains future-proof is by building a backend that allows for full programmability - where developers have the freedom to optimize and customize the system to fit their needs.
Lesson #2: Beware of ‘Best Practices’ That Don’t Scale
Scaling a live game isn’t just about handling more players; it’s about handling them correctly. Teemu recalls a war story that underscores this point. A game team enabled a database write-ahead log (WAL) - a best practice in web development - to ensure data integrity. But at the scale of a live game, this mechanism began choking the database, leading to a cascading failure. “Even AWS’s premium support wasn’t able to help,” Teemu notes. “We had to dig deep into the internals of the database implementation to find a fix.”
The takeaway? What works in traditional web applications doesn’t always translate to games. “Scaling databases for live games requires specialized knowledge,” Teemu warns. “Traditional solutions often don’t fit our world.”
Lesson #3: The Power of Vertical Integration
One of the biggest backend misconceptions is that success comes down to a single ‘killer feature.’ But as Teemu puts it, “It’s not about one thing - it’s about how everything works together.”
At Metaplay, backend features are deeply integrated into a cohesive system. Take error messages, for example: rather than just making them clearer, the team embedded them into dashboards, customer support tools, and infrastructure monitoring. Their command-line interface isn’t just a utility—it’s an entry point for logs, debugging, and automation.
“If you don’t think holistically about how all these elements interact, you’ll end up with a backend that feels like a collection of unrelated parts,” says Teemu. “That’s a recipe for friction, not scalability.”
Lesson #4: Networking Will Surprise You
For those used to building traditional web applications, networking in online games can be a rude awakening. Performance expectations are higher, failure cases are more extreme, and unpredictable constraints abound.
One of the unexpected hurdles Metaplay encountered was AWS’s lack of IPv6 support. “Players would boot up a game on Android, get assigned an IPv6-only address, and suddenly… nothing worked,” Teemu recalls. The solution? A custom tunneling system to ensure seamless connectivity. And that was just one of many surprises in a domain where mobile network instability, packet loss, and abrupt disconnections are the norm.
Thinking Beyond Today’s Problems
“A backend that works for a prototype isn’t necessarily suited for a live game,” Teemu points out. Too many studios start with quick-and-dirty backend solutions, only to realize too late that they need to rebuild.
To avoid this fate, Teemu offers some practical advice that newsletter subscribers can get directly to their inbox.
Final Thought: The Backend as an Enabler
After five years of solving some of the hardest problems in game backend development, Teemu sees one fundamental truth: “A backend is an enabler of long-term success - the best games need to be built on the best systems."
The right backend decisions - full programmability, vertical integration, observability, and resilience - can make or break a game’s ability to scale and thrive. And for those still on the journey, Teemu’s advice is simple: don’t build a backend for today, build one that can handle the unknowns of tomorrow.