Source: https://gameye.com/game-server-orchestration/ --- title: "Managed Game Server Orchestration Platform | Gameye" description: "Gameye is a managed game server orchestration platform. One REST API starts dedicated servers on bare metal in 0.5 seconds, then scales them and handles failover. No SDK, no Kubernetes." --- Platform # Managed game server orchestration Gameye is a managed game server orchestration platform. Your matchmaker calls one REST API, and Gameye starts a dedicated server in about 0.5 seconds and returns its IP and port. Gameye then scales your servers, moves new sessions away from failing providers, and shuts servers down when matches end. You add no SDK to your server and run no Kubernetes. Last modified October 25, 2026 ## Gameye at a glance - **Interface.** One REST call, `POST /session`, with a region and your image. No SDK in the server binary. - **Start time.** About 0.5 seconds from request to a running server, with its host IP and port returned to your matchmaker. - **Infrastructure.** Bare metal for your baseline, from 21 providers in 200+ datacenters, with cloud capacity added when demand exceeds it. - **Price.** $0.07 per vCPU per hour on demand, $0.035 reserved, billed per second. Egress is included. - **Reliability.** 99.7% contractual SLA, 99.9% measured uptime. More than 130 million sessions orchestrated. - **Engines.** Anything that builds a Linux dedicated server and runs in Docker. See [engine support](https://gameye.com/engine-support/). ## What orchestration does, and how Gameye does it Orchestration covers five jobs, from a match being ready to its server shutting down. Here is what Gameye does for each. | Job | How Gameye does it | | --- | --- | | Allocation | Your matchmaker calls `POST /session`. Gameye starts the container and returns the host IP and port in about 0.5 seconds. | | Placement | Gameye picks the node in the region you request, on bare metal first and cloud when the baseline is full. | | Scaling | There are no autoscaling rules to write. Warm pools keep pre-started containers in reserve per application and region for launch spikes, free weekends and tournaments. Warm containers are billable while held. | | Failover | Each region uses several providers. If one has a problem, new sessions go to healthy providers in the same region and existing sessions on healthy nodes keep running. | | Teardown | When the match ends, your backend stops the session with one API call and Gameye removes the container. Billing stops with it. | For the full request flow, see [how it works](https://gameye.com/how-it-works/) and the [API flow](https://gameye.com/platform-scope/). ## Gameye compared with other orchestration options Most studios choose between running Agones on their own Kubernetes cluster and using a managed service. Gameye is a managed service that also supplies the bare metal. | | Gameye | Agones (self-run) | Amazon GameLift Servers | Edgegap | | --- | --- | --- | --- | --- | | Who runs orchestration | Gameye | Your team, on Kubernetes | AWS | Edgegap | | Where servers run | Bare metal from 21 providers, cloud for peaks | Your cluster | AWS | Mixed edge pool of hyperscaler regions, commodity cloud and colocation | | Egress | Included | Your cloud's rates | Free on gen-6+ instances, per GB on gen-5 and older | Per GB on the public pool | Agones and GameLift both put calls into your server binary (Agones uses `sdk.Ready`, `sdk.Allocate` and `sdk.Shutdown`). Gameye needs none. Detailed comparisons: [Gameye vs Agones](https://gameye.com/gameye-vs-agones/), [Gameye vs GameLift](https://gameye.com/gameye-vs-aws-gamelift/), [Gameye vs Edgegap](https://gameye.com/gameye-vs-edgegap/). ## When Gameye fits, and when it doesn't ### Gameye fits when - You want dedicated servers without hiring for Kubernetes or on-call. - Your game is latency-sensitive and benefits from high-clock bare metal. - Your egress bill matters, or you want one price per vCPU-hour. - You expect launch spikes and want capacity staged in advance. ### Self-run Agones fits when - You already have a platform team running Kubernetes in production. - You need to customise the orchestration logic itself. - You want to run on hardware you already own or lease. ## Orchestration in production 250,000 concurrent players at launch [Chivalry 2](https://gameye.com/case-studies/chivalry-ii-tripwire-interactive/) by Torn Banner Studios scaled on launch day with zero downtime. 60%+ server cost reduction [Clone Drone in the Danger Zone](https://gameye.com/case-studies/clone-drone-in-the-danger-zone/) by Doborog Games, after moving to Gameye and dropping egress fees. 130M+ sessions orchestrated Across the platform. See [platform data and methodology](https://gameye.com/platform-data/). ## Frequently asked questions What is game server orchestration? Game server orchestration is the layer that starts dedicated game servers when a match is ready, places them close to players, scales capacity with demand, replaces servers that fail, and shuts them down when the match ends. It sits between your matchmaker and the machines your servers run on. Is Gameye a game server orchestration platform or a hosting provider? Gameye is a managed game server orchestration platform. It runs the orchestration for you and supplies the servers underneath it. That means bare metal from 21 providers, plus cloud capacity for peaks, so you don't rent or manage servers yourself. Do I need Kubernetes to use Gameye? No. Gameye replaces the Kubernetes and Agones setup you would otherwise run. Your matchmaker calls POST /session and Gameye handles placement, capacity, bursting and teardown. Do I need an SDK in my game server? No. Your server runs as a Docker container that starts, listens on a port and accepts player connections. Gameye manages it from outside through its REST API, so there are no Ready, Allocate or Shutdown calls to add to your binary. How much does Gameye cost? $0.07 per vCPU per hour on demand and $0.035 per vCPU per hour reserved, billed per second with no egress fees and no minimum session length. Which engines and matchmakers work with Gameye? Any dedicated server that builds for Linux and runs in a Docker container: Unreal Engine, Unity (including Mirror, FishNet and Photon Fusion), Godot and custom C++, Rust or Go servers. Gameye integrates with Pragma Engine, Nakama, FlexMatch, PlayFab and Unity Matchmaker, and any backend that can make an HTTP request. ## Go deeper Get started ## Start a Gameye Developer Trial. Sandbox testing with API access. Sign up with Google, push your Docker image and run your first session. [Start Building](https://trial.gameye.com/)