Choosing Cloud Models

A guide to game ops strategy

In the previous articles, we looked at how cloud changed the industry from an era of buying servers to an era of using only what is needed.

We also explored how that shift changed game operations in terms of scalability and agility.

At this point, business teams may naturally begin to ask:

“How do game companies actually use cloud in practice?”

“EC2, RDS, Unity Cloud, and Firebase… are all called cloud services, but what exactly is different about them?”

To answer that question, we need to understand that cloud is not just one technology.

It is divided into three operating models: IaaS, PaaS, and SaaS.

More recently, a fourth model has also become important: FaaS, or serverless.

ℹ️ The real difference between cloud models is not technical difficulty

Many people assume something like this:

“Is IaaS the difficult version of cloud, while SaaS is the easy one?”

But the real difference is not difficulty.

The real difference is this:

Who is responsible for what?

In other words, cloud is not simply a way to rent servers.

It is a way to choose the scope of operational responsibility.

ℹ️ Game Industry Examples of IaaS, PaaS, and SaaS

Even within a single game service, these models are usually mixed together.

A real time matchmaking server may be operated directly through IaaS

Ranking systems and log storage may rely on managed databases through PaaS

Build automation and collaboration tools may be used through SaaS

Intermittent functions, such as reward or coupon APIs that do not need to run all day, may use FaaS or serverless.

This means cloud is not one fixed method.

It is a combined operating structure.

ℹ️ Supercell’s cloud strategy

Supercell supports hundreds of millions of players worldwide with only three server engineers.

The reason is straightforward.

Instead of designing and operating everything directly, the company replaced as many components as possible with managed services based on PaaS and SaaS.

For example:

Player data storage uses managed databases with automatic backup

Battle balance and UI analysis use serverless functions and data pipelines.

Marketing and LiveOps analysis use SaaS based tools.

Operational dashboards are built as API based combined services.

In an official AWS session, Supercell explained its approach in simple terms:

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

When a company changes its cloud operating model, its team structure and priorities also change.

Instead of focusing on server management, the organization can focus more on content, design, and player experience.

ℹ️ Business questions that matter most

Cloud is not only a technical issue.

It is also a business decision.

That is why business teams do not need to manage servers directly, but they do need to understand how to evaluate the direction of the operating model.

1. Is our competitive advantage based on technical control or operational speed?

→ If direct control is the priority, IaaS may matter more.

If speed is more important, PaaS and SaaS should usually take a larger role.

2. Which areas truly need to be operated directly?

→ Routine maintenance work does not always need to stay in house.

Many repetitive tasks can be delegated to managed services.

3. Do we have a plan to change our infrastructure model as the game moves from launch to growth to stability?

→ In the early stage, PaaS is often useful because it allows faster setup.

In the stable stage, SaaS or other optimized service models may offer better cost efficiency.

4. Is this year’s KPI focused on technical internalization or operational efficiency?

→ The direction of the KPI affects the architecture choice and the cloud model as well.

These four questions help translate the language of technology into the language of management.

To understand cloud is ultimately to understand this:

What should we do ourselves, and what should we delegate?

♻️ Coming Next …

Cloud is no longer just an IT topic.

It is now a strategic choice that redesigns operational responsibility and resource allocation.

IaaS, PaaS, SaaS, and FaaS all represent different answers to one central question:

What should we build and manage ourselves, and what should we hand over to a provider?

In the next article, we will look at serverless, which sits at the far end of this trend.

Serverless does not mean servers disappear.

It means moving to a structure where you no longer need to manage them directly.

We will look at why the game industry is adopting this model so quickly, and what that means in real business and operational terms.

※ 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