The SDDC functionality is distributed across multiple clusters. A cluster can occupy one rack or multiple racks. The total number of racks for each cluster type depends on your scalability needs.
Figure 1. SDDC Cluster Architecture for a Single Availability Zone

Two racks host compute clusters of 19 ESXi hosts each. A third rack, with an external connection, hosts the shared compute and edge cluster (4 hosts) and the management cluster (4 hosts)
Figure 2. SDDC Cluster Architecture for Two Availability Zones
In environment with two availability zones in Region A, each zone spans over the management cluster and shared edge and compute cluster hosts on a single rack server. A zone represents a fault domain that is mapped to one rack.
Table 1. Design Decisions on Clusters and Racks

Decision ID

Design Decision

Design Justification

Design Implication

SDDC-PHY-004

In each availability zone, place the management cluster and the shared edge and compute cluster in the same rack.

The number of required compute resources for the management cluster (4 ESXi servers) and shared edge and compute cluster (4 ESXi servers) is low and does not justify a dedicated rack for each cluster.

You provide on-ramp and off-ramp connectivity to physical networks (for example, north-south Layer 3 routing on NSX Edge virtual appliances) to both the management and shared edge and compute clusters by using this management and edge rack.

Edge resources require external connectivity to physical network devices. Placing edge resources for management and tenants in the same rack minimizes VLAN spread.

The data centers must have sufficient power and cooling to operate the server equipment according to the selected vendor and products.

If the equipment in the entire rack fails, to reduce downtime associated with such an event, you must have a second availability zone or region.

SDDC-PHY-005

Allocate one or more racks to external storage.

To simplify the scale-out of the SDDC infrastructure, standardize the storage-to-racks relationship.

The storage system might arrive from the manufacturer in a dedicated rack or set of racks. In this way, a storage system of this type is accommodated for in the design.

The data centers must have sufficient power and cooling to operate the storage equipment according to the selected vendor and products.

SDDC-PHY-006

Use two separate power feeds for each rack.

Redundant power feeds increase availability by ensuring that failure of a power feed does not bring down all equipment in a rack.

Combined with redundant network connections to a rack and in a rack, redundant power feeds prevent a failure of the equipment in an entire rack.

All equipment used must support two separate power feeds. The equipment must keep running if one power feed fails.

If the equipment of an entire rack fails, the cause, such as flooding or an earthquake, also affects neighboring racks. Use a second region to reduce downtime associated with such an event.

SDDC-PHY-007

For each availability zone, mount the compute resources (minimum of 4 ESXi hosts) for the management cluster together in a rack

Mounting the compute resources for the management cluster together can ease physical data center design, deployment, and troubleshooting.

None.

SDDC-PHY-008

For each availability zone, mount the compute resources for the shared edge and compute cluster (minimum of 4 ESXi hosts) together in a rack.

Mounting the compute resources for the shared edge and compute cluster together can ease physical data center design, deployment, and troubleshooting.

None.