This document explains the deployment hierarchy for
Google Distributed Cloud (GDC) air-gapped, which includes universes, regions, and zones. To
provide high availability for your air-gapped resources and infrastructure, you
must understand the locations that host your environment and the strategic
boundaries that enforce available configurations for your applications.

This document is for the following audience groups:

- Infrastructure operators who are responsible for deploying and managing GDC zones and global infrastructure.
- Platform administrators, such as network and IT administrators, who are responsible for configuring multi-zone services.
- Application operators who are responsible for developing and managing applications in a GDC universe.

For more information, see
[Audiences for GDC air-gapped documentation](https://docs.cloud.google.com/distributed-cloud/hosted/docs/latest/gdcag/resources/audiences.md.txt).

## Deployment hierarchy overview

A GDC air-gapped deployment follows a hierarchical
structure consisting of universes, regions, and zones.

A *universe* is a high-level grouping of interconnected zones that share network
connectivity and a control plane. Even though zones can be geographically
distant, a *regional boundary* is established when distinct zones are physically
close enough to have low-latency communication.

A *zone* is the building block of your GDC air-gapped
deployment. Each zone is a self-contained implementation that doesn't require
connectivity to Google Cloud at any time. A zone manages its own infrastructure,
services, APIs, and tooling through a local control plane.

For example, the following diagram illustrates a GDC
multi-zone universe:

![A universe consists of zones that are grouped across regions.](https://docs.cloud.google.com/static/distributed-cloud/hosted/docs/latest/gdcag/images/gdc-universe.png)

This diagram has two regions, each existing within a single universe. Region 1
is highlighted to show three zones that each operate various zonal services,
such as virtual machines (VM) and container applications. There are also global
services that span all zones and regions in the universe, such as
Identity and Access Management (IAM) and load balancing.

For example, you could have a GDC universe that spans
across far-reaching geographic regions. Consider Region 1 existing in Virginia,
USA with three zones:

- Zone A, such as `us-virginia1-a`
- Zone B, such as `us-virginia1-b`
- Zone C, such as `us-virginia1-c`

Likewise, you could have Region 2 reside in the Netherlands and include zones in
Amsterdam:

- Zone A, such as `eu-ams1-a`
- Zone B, such as `eu-ams1-b`
- Zone C, such as `eu-ams1-c`

Zones within a region must be geographically close to each other, but regions
can span great distances. For example, a zone might have a fiber distance
maximum limit of 50 km to another zone within the same region, but the
region can span 100s of km from another region. Contact your infrastructure
operator group for more information on your specific latency requirements.

### About zones

A *zone* contains four layers that can operate in isolation without requiring
connectivity to Google Cloud or the public internet:

- **Hardware**: the foundational equipment fully integrated into the racks of your data center. This layer provides the compute, storage, network, and GPU capacity for your system and integrated workloads.
- **Infrastructure**: the underlying hardware management system. This layer provides interfaces that enable the software to run without needing hardware specific configurations.
- **Service Platform**: the framework for building services on GDC. This layer provides consistency among managed services and marketplace services.
- **Managed and Marketplace services**: the customer-facing cloud services running on GDC. This layer provides a consistent implementation of the data plane for workloads and services.

These components give you the flexibility to adjust your zonal configuration to
meet your unique availability and resilience requirements. Understanding this
structure gives the foundation for understanding the common air-gapped zone
configurations.

## Deployment configurations

The following sections describe common deployment configurations that you can
use for GDC air-gapped.

### Single zone configuration

A *single zone* configuration lets you deploy a universe where one zone acts as
the sole region. This configuration is possible because each zone operates
independently. You'll typically use this setup for initial deployments or for
workloads that don't require high availability or robust disaster recovery.

The main limitation for a single zone universe is that it creates a single point
of failure, making your system vulnerable to outages. A single point of failure
happens when you deploy resources exclusively to a single zone.
[Zonal resources](https://docs.cloud.google.com/distributed-cloud/hosted/docs/latest/gdcag/resources/multi-zone/api-servers.md.txt#resource-types)
are impacted as a group when an outage or service disruption occurs in that
specific zone.

### Multi-zone configuration

A *multi-zone configuration* lets you strategically distribute your applications
and services across several zones. This setup improves your system's resilience
and fault tolerance by providing geographically distributed resources.

This configuration also provides *high availability* , which lets your
application or service remain continuously accessible and operational even if a
zone experiences an outage. You achieve high availability through redundancy by
deploying and managing resources across multiple zones within a universe.
GDC offers a
[subset of services](https://docs.cloud.google.com/distributed-cloud/hosted/docs/latest/gdcag/reference/multi-zone-services.md.txt) that provide
global capabilities so the resource is managed across multiple zones in a
universe by default without manual replication.
[Global resources](https://docs.cloud.google.com/distributed-cloud/hosted/docs/latest/gdcag/resources/multi-zone/api-servers.md.txt#resource-types)
span multiple zones by default, which makes them available even during a zonal
outage.

A key benefit to a multi-zone configuration is that each zone can function as a
distinct disaster domain. A *disaster domain* is a group of resources
susceptible to simultaneous failure due to their shared physical location or
dependency. By strategically positioning zones far enough apart, your system
becomes more resilient against widespread failures.

*Zonal asymmetry* allows for organization and hardware configurations to be
different across zones in a GDC universe. Zonal
asymmetry is a [Preview](https://docs.cloud.google.com/distributed-cloud/hosted/docs/latest/gdcag/resources/feature-stages.md.txt#preview) feature
starting in version 1.14.4.

## Key capabilities of a multi-zone configuration

Multi-zone configurations let you establish resilient architecture because each
zone can function as distinct disaster domains. Distributing resources across
these independent disaster domains lets you enhance your air-gapped systems with
the key capabilities outlined in the following sections.

### High availability

You can create highly available applications to provide durable services that
are able to withstand outages. High availability strategies, such as load
balancing and data replication, transform fragile zonal applications into
resilient and reliable global services. For more information, see
[High availability for your apps](https://docs.cloud.google.com/distributed-cloud/hosted/docs/latest/gdcag/platform-application/pa-ao-operations/ha-apps/overview.md.txt).

### Data protection

You can use a multi-zone configuration to keep your data secure and accessible
even in the event of an outage using data protection strategies that let you set
up data replication to create geographically separated backups. For more
information, see
[Data protection with multi-zone storage](https://docs.cloud.google.com/distributed-cloud/hosted/docs/latest/gdcag/resources/multi-zone/data-protection.md.txt).

### Network traffic management

Building a durable network is essential to creating a highly available
GDC system. You can configure networking strategies
across multiple zones to plan for failover procedures during a potential zonal
outage. Balancing your network traffic across zones can also positively impact
application performance. For more information, see
[Network traffic management for multiple zones](https://docs.cloud.google.com/distributed-cloud/hosted/docs/latest/gdcag/resources/multi-zone/networking.md.txt).

### Permissions control

You can implement cross-zone roles to apply access control strategies on a
global scale, which removes zone-specific permissions management that can be
tedious and difficult to manage. Likewise, you can seamlessly manage
authentication across zones for your human users and services with global access
accounts. For more information, see
[Permissions control for a multi-zone universe](https://docs.cloud.google.com/distributed-cloud/hosted/docs/latest/gdcag/resources/multi-zone/permissions.md.txt).

## Limitations

GDC air-gapped universes have the following limitations:

- A universe with two zones can't provide an automated recovery strategy, regardless of region configuration, and must be recovered manually. Universes with three or more zones can trigger recovery steps automatically.
- GDC provides regions only as a grouping concept, and doesn't provide regional services. Regions in a GDC universe are provided to closely align with public Google Cloud architecture, and for future development and planning.

## What's next

- Learn about the [global and zonal resource types](https://docs.cloud.google.com/distributed-cloud/hosted/docs/latest/gdcag/resources/multi-zone/api-servers.md.txt) available in a GDC universe.
- Explore the [High availability guide](https://docs.cloud.google.com/distributed-cloud/hosted/docs/latest/gdcag/platform-application/pa-ao-operations/ha-apps/overview.md.txt) to ensure your application is resilient to local zone failures.
- Read the [Resource hierarchy](https://docs.cloud.google.com/distributed-cloud/hosted/docs/latest/gdcag/resources/resource-hierarchy.md.txt) document for more information on how resources are managed hierarchically within a zone.