VMware Cloud Foundation supports multiple topologies that provide different levels of availability and scale.

Availability Zones and VMware Cloud Foundation Instances

Availability zone

An availability zone is a fault domain at the SDDC level.

You create multiple availability zones for the purpose of creating vSAN stretched clusters. Using multiple availability zones can improve availability of management components and workloads running within the SDDC, minimize downtime of services, and improve SLAs.

Availability zones are typically located either within the same data center, but in different racks, chassis, rooms, or in different data centers with low-latency high-speed links connecting them.

Note:

Only stretched clusters created by using the Stretch Cluster API, and are therefore vSAN storage based, are considered by and treated as stretched clusters by VMware Cloud Foundation.

VMware Cloud Foundation Instance

Each VMware Cloud Foundation instance is a separate VMware Cloud Foundation deployment and might contain one or two availability zones. VMware Cloud Foundation instances may be geographically separate.

VMware Cloud Foundation Topologies

Several topologies of VMware Cloud Foundation exist according to the number of availability zones and VMware Cloud Foundation instances.

Table 1. VMware Cloud Foundation Topologies

Topology

Description

Single Instance - Single Availability Zone

Workload domains are deployed in a single availability zone.

Single Instance - Multiple Availability Zones

Workload domains might be stretched between two availability zones.

Multiple Instances - Single Availability Zone per VMware Cloud Foundation instance

Workload domains in each instance are deployed in a single availability zone.

Multiple Instances - Multiple Availability Zones per VMware Cloud Foundation instance

Workload domains in each instance might be stretched between two availability zones.

Figure 1. Choosing a VMware Cloud Foundation Topology
The first decision is for disaster recovery or more than 24 VI workload domains. If you need them, you select multiple instances with a single or with multiple availability zones. If you do not need them, you select a single instance with a single or multiple availability zones.

Single Instance - Single Availability Zone

Single Instance - Single Availability Zone is the simplest VMware Cloud Foundation topology where workload domains are deployed in a single availability zone.

The Single Instance - Single Availability Zone topology relies on vSphere HA to protect workloads against host failures.

Figure 2. Single VMware Cloud Foundation Instance with a Single Availability Zone

When using multiple availability zones, the management domain must be stretched between availability zones. VI workload domains can be stretched between the two zones or run only in one zone.
Table 2. Single Instance - Single Availability Zone Attributes

Attributes

Detail

Data centers

Single data center

Workload domain cluster rack mapping

  • Workload domain cluster in a single rack

  • Workload domain cluster spanning multiple racks

Scale

  • Up to 25 workload domains

  • Up to 15 workload domains in a single vCenter Single Sign-On domain including the management domain.

Resilience

vSphere HA protects against host failures

Single Instance - Multiple Availability Zones

You protect your VMware Cloud Foundation environment against a failure of a single hardware fault domain by implementing multiple availability zones.

Incorporating multiple availability zones in your design can help reduce the blast radius of a failure and can increase application availability. You usually deploy multiple availability zones across two independent data centers.

Figure 3. Single VMware Cloud Foundation Instance with Multiple Availability Zones

When using multiple availability zones, the management domain must be stretched between availability zones. VI workload domains can be stretched between the two zones or run only in one zone.
Table 3. Single Instance - Multiple Availability Zone Attributes

Attributes

Detail

Workload domain cluster rack mapping

  • Workload domain cluster in a single rack.

  • Workload domain cluster spanning multiple racks.

  • Workload domain cluster with multiple availability zones, each zone in a single rack.

  • Workload domain cluster with multiple availability zones, each zone spanning multiple racks (not supported with a multi-rack compute cluster).

Stretched cluster

  • Because availability zones use VMware vSAN™ stretched clusters, the bandwidth between the zones must be at least 10 Gbps and the round-trip latency must be less than 5 ms.

  • Having the management domain on a vSAN stretched cluster is a prerequisite to configure and implement vSAN stretched clusters in your VI workload domains. In this way, the critical management components to manage the VI workload domains are still available if a site failure occurs.

  • You can have up to two availability zones.

Scale

  • Up to 25 workload domains.

    Up to 15 workload domains in a single vCenter Single Sign-On domain.

Resilience

  • vSphere HA protects workloads against host failures.

  • Multiple availability zones protect against data center failures.

Multiple Instances - Single Availability Zone per Instance

You protect against a failure of a single VMware Cloud Foundation instance by implementing multiple VMware Cloud Foundation instances.

Incorporating multiple VMware Cloud Foundation instances in your design can help reduce the blast radius of a failure and can increase application availability across larger geographical distances than cannot be achieved by using multiple availability zones. You usually deploy this topology in the same data center for scale or across independent data centers for resilience.

Figure 4. Multiple VMware Cloud Foundation Instances with a Single Availability Zone per Instance

When using multiple instances, each instance contains a management domain and optionally, VI workload domains.
Table 4. Multiple Instance - Single Availability Zone Attributes

Attributes

Detail

Workload domain cluster rack mapping

  • Workload domain cluster in a single rack

  • Workload domain cluster spanning multiple racks

Multiple instances

Using multiple VMware Cloud Foundation instances can facilitate the following use cases:

  • Disaster recovery across different VMware Cloud Foundation instances at longer distances

  • Scale beyond the maximums of a single VMware Cloud Foundation instance.

  • Co-location of end users and resources

If you plan to use NSX Federation between VMware Cloud Foundation instances, the following considerations exist:

  • Up to four locations when using medium-size NSX Global Managers

  • Up to 16 locations when using large or extra large NSX Global Managers

  • Up to four locations per cross-instance Tier-0 gateway

  • Check https://configmax.vmware.com/ to select the right NSX Global Managers form factor for your scale needs.

  • Lifecycle management must be planned carefully

Scale

  • Up to 25 workload domains per VMware Cloud Foundation instance

    Up to 15 workload domains in a single vCenter Single Sign-on domain per instance

Resilience

  • vSphere HA protects workloads against host failures.

  • Deploying multiple instances can protect against natural disasters by providing recovery locations at greater geographical distances.

Multiple Instances - Multiple Availability Zones per Instance

You protect against a failure of a single VMware Cloud Foundation instance by implementing multiple VMware Cloud Foundation instances. Implementing multiple availability zones in an instance protects against a failure of a single hardware fault domain.

Incorporating multiple VMware Cloud Foundation instances into your design can help reduce the blast radius of a failure and can increase application availability across larger geographical distances that cannot be achieved using multiple availability zones.

Figure 5. Multiple VMware Cloud Foundation Instances with Multiple Availability Zones per Instance

Multiple instances, each with the management domain stretched between 2 availability zones, and optionally VI workload domains in a single availability zone, or spanning multiple availability zones.
Table 5. Multiple Instance - Multiple Availability Zone Attributes

Attributes

Detail

Workload domain cluster rack mapping

  • Workload domain cluster in a single rack

  • Workload domain cluster spanning multiple racks

  • Workload domain cluster with multiple availability zones, each zone in a single rack

  • Workload domain cluster with multiple availability zones, each zone spanning multiple racks (not supported with multi-rack compute clusters)

Multiple instances

Using multiple VMware Cloud Foundation instances can facilitate the following:

  • Disaster recovery across different VMware Cloud Foundation instances at longer distances

  • Scale beyond the maximums of a single VMware Cloud Foundation instance

  • Co-location of end users and resources

If you plan to use NSX Federation between instances, VMware Cloud Foundation consider the following:

  • Up to four locations when using medium-size NSX Global Managers

  • Up to 16 locations when using large-size NSX Global Managers

  • Up to four locations per stretched Tier-0 gateway

  • Lifecycle management must be planned carefully

Stretched cluster

  • Because availability zones use VMware vSAN™ stretched clusters, the bandwidth between the zones must be at least 10 Gbps and the round-trip latency must be less than 5 ms.

  • You can have up to two availability zones.

  • Having the management domain on a vSAN stretched cluster is a prerequisite to configure and implement vSAN stretched clusters in your VI workload domains.

Scale

  • Up to 25 workload domains per VMware Cloud Foundation instance

    Up to 15 workload domains in a single vCenter Single Sign-On domain per instance

Resilience

  • vSphere HA protects workloads against host failures.

  • Multiple availability zones protect against data center failures.

  • Multiple instances can protect against natural disasters by providing recovery locations at greater geographical distances.