The Serverless Mindset

From operations to execution

So far, we have looked at how clouds are divided into IaaS, PaaS, and SaaS, and how game operations have shifted from building everything directly to delegating and combining services.

At this point, a new question may come up.

“Is serverless the next step?”

“Does it really mean servers disappear?”

The answer is no. Serverless does not mean there are no servers.

Serverless means a structure in which servers no longer need to be managed directly. In other words, the responsibility for operations moves from people to the cloud.

ℹ️ The Core of Serverless: Execution Over Operations

In the past, people had to launch servers, configure networks, deploy services, monitor systems, scale resources, and recover from failures.

In a serverless model, these tasks run in an automated environment. The operator only needs to define when and under what conditions a task should run.

So operations have not disappeared. The ownership of operations has changed. It has moved to the cloud.

As a result, game companies can spend less time on infrastructure management and more time on player experience and content quality.

ℹ️ How serverless works?

Serverless follows an event driven structure. It runs only when a specific event happens.

For example:

When a player logs in

When payment is completed

When an event reward is granted

Whenever the system needs to perform a new action, a function runs only at that moment. Once the task is finished, it stops immediately.

If there is no usage, there is no charge.

That is why serverless is often seen as a cost-efficient operating model that also handles variable traffic well.

ℹ️ How serverless is used in games?

Serverless is especially strong in short term traffic spikes and event-based features.

In other words, serverless is a smart form of infrastructure that works only when needed.

It improves operational efficiency and also gives teams an environment where they can experiment quickly as market conditions change.

ℹ️ Apex Legends: a serverless style approach

EA’s Apex Legends is a large scale multiplayer game that serves tens of millions of players worldwide.

In the early stage, the team operated servers directly. But traffic spikes during global events and seasonal updates created a major burden. To solve this, Respawn chose to migrate to Amazon GameLift.

Amazon GameLift is not fully serverless, but it is a form of managed infrastructure with built in auto scaling and recovery features.

AWS and its partners worked with Respawn to design a gradual migration plan with low risk. They minimized code changes and created conditions that allowed a stable transition without service interruption or large scale QA.

Traffic was moved region by region, starting from smaller markets and verifying stability step by step. In the end, millions of players were moved to the new infrastructure in just ten days.

The results were clear.

Average global latency fell by 30%, and in some regions the improvement reached 50%. Packet loss also declined by 5% on average worldwide, improving overall service quality.

Also, the previous hardware setup, which had been divided across fifteen different configurations, was simplified into just two. At the same time, matchmaking logic and operational efficiency improved significantly. Long standing issues such as rubber

After the migration, Respawn explained the result in simple terms:

“Now we focus on player experience, not servers.”

This was not a model where servers disappeared completely. But it clearly showed that an operating approach built around not managing servers directly can improve stability, cost efficiency, and speed of operations at the same time.

* Rubber banding refers to a situation where a character appears to move forward in game, then suddenly snaps back to a previous position. It usually happens because of network delay or data synchronization issues between the server and the client.

ℹ️ Riot Games: focusing on player experience over operations

Riot Games runs titles such as League of Legends and VALORANT as global live services and must handle large-scale real-time traffic.

To support match data analysis, replay processing, and global event response, Riot adopted serverless architecture using services such as AWS Lambda.

When traffic rises sharply in a specific region, functions run only when needed. When traffic falls, they stop immediately.

Riot Engineering described the goal clearly:

“We no longer think about adding more servers. Our operational goal is to maintain the quality of player experience.”

Through this serverless mindset, Riot has built a more agile operating model centered on automation, scaling, and analysis.

ℹ️ The philosophical starting point: “We do not manage servers”

One of the earliest companies to put this philosophy into practice was Supercell.

“We do not manage servers. We focus on games.”

— Supercell Tech Talk (AWS re:Invent)

Today, companies such as EA and Riot are expanding the same philosophy at much larger scale.

It is also helpful to understand the difference between managed infrastructure and managed services.

Managed infrastructure refers to a structure where the platform manages servers or networks on behalf of the user. Examples include RDS and EC2 Auto Scaling.

→ this is similar to a service that regularly replaces the filter in a water purifier.

Managed service refers to a structure where the platform automatically handles a specific function. Examples include Lambda and Firebase Functions.

→ this is similar to a kitchen system that automatically starts cooking when an order comes in

ℹ️ Key questions business teams should ask about serverless

Serverless is not the right answer for every game.

Business teams should ask the following questions to decide whether it fits their service.

Is our traffic stable, or does it fluctuate heavily?

→ If traffic changes a lot, serverless may be more efficient.

Is the core strength of our operations team infrastructure management, or content operations?

→ If the answer is content operations, a managed or serverless structure may be more suitable.

How short is our development and deployment cycle?

→ If releases and testing happen frequently, automated scaling becomes more important.

Is our cost structure built around fixed cost or usage-based cost?

→ Serverless creates a more flexible model built around operating expense.

These questions are not really asking, “Should we remove servers?” They are asking, “Who should own operations?”

♻️ Coming Next …

Serverless is not a technology that removes servers.

It is a way of thinking that automates operations and shifts the focus toward execution.

Today, the focus of game operations is no longer only about how stably systems can be maintained. It is increasingly about how quickly teams can respond.

In that environment, serverless is more than a technical model. It is a structural innovation that helps game businesses become more agile.

In the next article, we will look at how data is handled on the cloud, and how data pipelines and real time analytics are changing decision making in game business.

※ Disclaimer: This content reflects the author’s personal views and includes only publicly available examples. It does not represent the official position of any company mentioned