Your studio doesn’t plan to spend months on infrastructure. But your backend team has other ideas.

Someone on your engineering team, probably several people, are managing a Kubernetes cluster instead of pushing feature updates. After all, somebody has to wire it all up, and it usually ends up being your best engineers.

With Armadas, we've built out the orchestration layer that keeps your team on the game and not your infrastructure.

What Armadas Do

An Armada is a regional cluster of identical, ephemeral game servers, all running the same build. You commit the single build along with your configurations and GameFabric will manage your deployment, scaling, health checks, and matchmaking handoff to the letter. Servers spin up before they're needed, sit in a ready pool, get allocated to a match, run for the duration of the session, and shut down. Your Armada will replenish the pool automatically.

GameFabric can provision over 250 game servers every second, so your infrastructure keeps pace with matchmaking demand on launch day, content releases, and unexpected traffic spikes.

If you’re running across multiple regions, ArmadaSets extend this to a master template that replicates your infrastructure globally. One deployment update syncs across every active region, while maintaining isolated, region-by-region version histories.

Hybrid, Out of the Box

Each Armada can run across both bare metal and cloud in the same region, with priority-based scheduling. Your bare metal absorbs your predictable baseline load first, while cloud covers your peaks. When demand drops, cloud scales back. It manages this automatically, without custom logic on your side.

This is the architecture a lot of studios try to build for themselves after one too many runaway cloud bills. With Armadas, you get it complete from day one.

Patch Day Without the Stress

Shipping a patch to a live player base is inevitably stressful, but it’s doubly so when you don't have an instant rollback path. Armadas have revisions: every configuration change creates a versioned snapshot of your state. When you push a new build, you can roll it out gradually or run old and new versions in parallel while you validate. If something goes wrong, rolling back is a deliberate action, not an emergency scramble.

If you've had a bad patch land on live players and had to figure it out on the fly, this is for you.

Matchmaking in One Call

Hooking a matchmaker up to a constantly shifting server is notoriously finicky. So GameFabric offers an Allocator service, which handles the handshake between your matchmaker and live servers. When a match is ready, your matchmaker sends a single API call to the Allocator. It picks an available server, and immediately passes along any match-specific data like map, mode, or player list.

We’re compatible with standard matchmaking patterns and custom matchmakers via API, so you get to keep the matchmaker you're most comfortable with.

What Your Team Gets Back

The hours your team spend running Kubernetes, wrangling Agones, and maintaining custom  tooling are a real opportunity cost. But none of it makes your game better. Your infrastructure will keep demanding more attention as your team and player base grow.

Armadas are built on Agones and Kubernetes, so engineers who know those stacks aren't starting over. They just aren't responsible for operating them anymore. At the end of the day, you’re not a Kubernetes company that happens to make games; you’re a game studio that needs a toolset built to serve you.

Want to see what Armadas look like for your setup? Explore our documentation or book a demo to walk through your studio's setup with our team.