Regions, Sites & Locations

GameFabric organizes compute capacity into three layers. Understanding how they relate to each other helps you configure deployments correctly and diagnose capacity or scheduling issues.

Locations and Sites are global and provisioned by the platform operator (or via GameFabric Cloud self-service). Regions are created and managed by your team inside a specific Environment. A Region does not create capacity — it selects and organizes existing capacity for your deployments.

See Capacity Types for details on bare metal, GameFabric Cloud, and BYOC capacity.

Regions

A Region is the capacity target for your game server deployments. When you create an Armada or Formation, you target a Region. GameFabric schedules game server processes onto the Sites that back that Region.

What counts as a logical region is a game-design decision, not a geographical one. You might group Frankfurt, London, and Amsterdam into a single "EU" Region if your matchmaker treats Europe as one pool. Or you might define "EU-West" and "EU-North" separately if latency matters for your title. Region Types add a further dimension: you can prioritize bare metal Locations over cloud within the same Region to manage cost, without changing any deployment configuration.

Region Types

Every Region has one or more Types. A Type defines a class of infrastructure within the Region and maps it to a set of Locations.

Common examples:

• baremetal — maps to Locations backed by Nitrado bare metal hardware
• cloud — maps to Locations backed by GameFabric Cloud or BYOC capacity

Types have a priority (0 = highest priority, 1 = lower, and so on). GameFabric fills the highest-priority type first. Use this to prefer lower-cost bare metal capacity and only overflow to cloud when bare metal is exhausted.

Each Type also carries an optional template that applies to all game servers scheduled in that Type:

• Environment variables — injected into every game server container in this Type. Useful for passing infrastructure context (e.g. region name, capacity tier) without hardcoding it in your image.
• Scheduling strategy — controls how game servers are placed across nodes:
  • Packed (default) — game servers are co-located on shared nodes. Maximizes density.
  • Distributed — game servers spread across nodes. Reduces blast radius if a node fails.

Capacity tracking

The Regions list shows aggregate CPU and memory usage across all Types for each Region as progress bars. These bars reflect the combined used and limit values across all Types.

Protection status

When SteelShield™ mitigations are active on your account, the Regions list shows a Protection Status column. This status is derived from the Sites that back the Region's Locations:

Status	Meaning
Protected	All backing Sites are protected
Partially Protected	Some backing Sites are protected
Unprotected	No backing Sites are protected
Unknown	Protection state cannot be determined (no Sites resolved, or SteelShield not active)

Managing Regions

The following operations are available from the Regions list:

• Edit — opens a two-tab modal. The General tab updates display name and description. The Types tab adds, removes, reorders, and reconfigures types — including location assignments, priority, environment variables, and scheduling strategy.
• Clone — creates a new Region with the same type configuration. Useful when setting up parallel environments (e.g. cloning a prod region to create staging).
• Delete — removes the Region. Running deployments targeting this Region should be stopped first.

Region name constraints

• Maximum 24 characters
• Lowercase alphanumeric, . and - only
• Must start and end with a lowercase letter or digit

For a step-by-step walkthrough of creating a Region, see Set up your environment.

Filtering

The Regions list can be filtered by protection status and by type name. Use the type filter to quickly find all Regions of a specific capacity class (for example, all Regions with a cloud type to check overflow capacity).

Locations

A Location is a geographic node that groups one or more Sites. It is the unit that Region Types reference when targeting capacity.

Locations are global and read-only for users, with two exceptions:

• Users with the clouds: post capability can request GameFabric Cloud capacity at an eligible Location.
• Users with the clouds: patch,delete capability can remove GameFabric Cloud capacity from a provisioned Location.

Location types

Locations carry a type label (g8c.io/location-type) that identifies the kind of capacity they represent:

Type	Meaning
baremetal	Nitrado-owned physical hardware
managed	GameFabric Cloud (GCP-provisioned, billed through GameFabric)
byoc	Bring Your Own Cloud (customer's cloud account)
unknown	Type has not been classified

GameFabric Cloud Locations display a Managed tag in the Locations list.

Default filters

The Locations list opens with default filters applied to reduce noise:

• Sites: used and unused (excludes unprovisioned)
• Regions: assigned (excludes Locations not referenced by any Region)

Clear these filters to see all Locations including those with no active capacity.

Requesting GameFabric Cloud capacity

On a Location that supports GameFabric Cloud provisioning, a Request Cloud Location button appears in the actions column. Clicking it opens a form to request capacity at that Location. While provisioning is in progress the row shows a Provisioning requested badge. Once complete, the Location's type becomes managed and Sites appear under it.

To deprovision previously requested capacity, click Remove Cloud Location on a managed Location. The row shows Deprovisioning requested while the operation is in progress.

INFO

GameFabric Cloud capacity cannot be added to Locations that already have BYOC capacity.

Cross-references in the Locations view

• The Sites count on each row opens the Sites list filtered to that Location.
• The Associated Regions count shows how many Regions across all environments reference this Location. Click it to see the list with links to each environment's Regions page.

Filtering

The Locations list can be filtered by site state, region assignment, and location type. When diagnosing why a Region has no available capacity, start here — check whether the Locations referenced by your Region's Types have active, non-cordoned Sites.

Sites

A Site is a capacity cluster — a set of physical or virtual nodes represented as a Kubernetes namespace — that belongs to a Location. Sites are read-only in the UI. They are created and managed by the platform operator, or provisioned automatically when you request GameFabric Cloud capacity.

Site states

State	Meaning
Connected	The Site is healthy and accepting new game server scheduling
Pending	The Site is being provisioned and is not yet ready
Terminating	The Site is being decommissioned
Error	The Site has a connection or configuration problem; the reason is shown in the details

Cordoned Sites

A cordoned Site does not accept new game server scheduling. Running game servers continue until they shut down naturally, but new ones are not started on that Site. The Sites list marks cordoned Sites with a Cordoned tag.

Cordoning is an operator action. It is typically applied when a Site is being prepared for maintenance or deprovisioning, or when a freshly provisioned Site should not receive traffic yet.

Capacity per Site

The Sites list shows:

• Total CPU and Total Memory — the configured resource limit for the Site
• CPU used and Memory used — live bars showing current utilization

Protection status

When SteelShield™ is active, the Sites list adds a Protection Status column showing the per-Site protection state: Protected, Partially Protected, Unprotected, or Unknown. Per-Site protection states are the source of truth for the Region-level protection status.

Site details

Click Details on any Site row to view site-level environment variables configured by the operator. These variables are injected into all game servers scheduled on that Site.

Filtering

The Sites list can be filtered by state and protection status. Filter by cordoned to audit Sites that are not accepting new scheduling.