vSphere Cluster Design for VMware Cloud Foundation

vSphere cluster design must consider the requirements for standard, stretched and remote clusters and for the life cycle management of the ESXi hosts in the clusters according to the characteristics of the workloads.

Logical vSphere Cluster Design for VMware Cloud Foundation

The cluster design must consider the characteristics of the workloads that are deployed in the cluster.

When you design the cluster layout in vSphere, consider the following guidelines:

• Compare the capital costs of purchasing fewer, larger ESXi hosts with the costs of purchasing more, smaller ESXi hosts. Costs vary between vendors and models. Evaluate the risk of losing one larger host in a scaled-up cluster and the impact on the business with the higher chance of losing one or more smaller hosts in a scale-out cluster.

• Evaluate the operational costs of managing a few ESXi hosts with the costs of managing more ESXi hosts.

• Consider the purpose of the cluster.

• Consider the total number of ESXi hosts and cluster limits.

Figure 1. Logical vSphere Cluster Layout with a Single Availability Zone for VMware Cloud Foundation

Figure 2. Logical vSphere Cluster Layout for Multiple Availability Zones for VMware Cloud Foundation

Remote Cluster Design Considerations

Remote clusters are managed by the management infrastructure at the central site.

Table 1. Remote Cluster Design Considerations

Remote Cluster Attribute	Consideration
Number of hosts per remote cluster	Minimum: 3 Maximum: 16
Number of remote clusters per VMware Cloud Foundation instance	Maximum: 8
Number of remote clusters per VI workload domain	1
Cluster types per VI workload domain	A VI workload domain can include either local clusters or a remote cluster.
Latency between the central site and the remote site	Maximum: 100 ms
Bandwidth between the central site and the remote site	Minimum: 10 Mbps

vSphere Cluster Life Cycle Method Design for VMware Cloud Foundation

vSphere Lifecycle Manager is used to manage the vSphere clusters in each VI workload domain.

When you deploy a workload domain, you choose a vSphere cluster life cycle method according to your requirements.

Table 2. vSphere Lifecycle Manager choices

Cluster Life Cycle Method	Description	Benefits	Drawbacks
vSphere Lifecycle Manager images	vSphere Lifecycle Manager images contain base images, vendor add-ons, firmware, and drivers.	Supports vSAN stretched clusters. Supports VI workload domains with vSphere with Tanzu. Supports NVIDIA GPU-enabled clusters. Supports 2-node NFS, FC, or vVols clusters.	An initial cluster image is required during workload domain or cluster deployment.
vSphere Lifecycle Manager baselines	An upgrade baseline contains the ESXi image and a patch baseline contains the respective patches for ESXi host.	Supports vSAN stretched clusters. Supports VI workload domains with vSphere with Tanzu.	Not supported for NVIDIA GPU-enabled clusters. Not supported for 2-node NFS, FC, or vVols clusters.

vSphere Cluster Design Requirements and Recommendations for VMware Cloud Foundation

The design of a vSphere cluster is a subject to a minimum number of hosts, design requirements, and design recommendations.

For vSAN design requirements and recommendations, see vSAN Design Requirements and Recommendations for VMware Cloud Foundation.

The requirements for the ESXi hosts in a workload domain in VMware Cloud Foundation are related to the system requirements of the workloads hosted in the domain. The ESXi requirements include number, server configuration, amount of hardware resources, networking, and certificate management. Similar best practices help you design optimal environment operation

vSphere Cluster Design Considerations

You consider different number of hosts per cluster according to the storage type and specific resource requirements for standard and stretched vSAN clusters.

Table 3. Host-Related Design Considerations per Cluster

Attribute	Specification	Management Domain (Default Cluster)	Management Domain (Additional Clusters) or VI Workload Domain (All Clusters)
Minimum number of ESXi hosts	vSAN (single availability zone)	4	3
	vSAN (two availability zones)	8	6
	NFS, FC, or vVols	Not supported	2 VI workload domain only Requires vSphere Lifecycle Manager images 3 Additional management clusters
Reserved capacity for handling ESXi host failures per cluster	Single availability zone	25% CPU and memory Tolerates one host failure	33% CPU and memory Tolerates one host failure
	Two availability zones	50% CPU and memory Tolerates one availability zone failure	50% CPU and memory Tolerates one availability zone failure

vSphere Cluster Design Requirements VMware Cloud Foundation

You must meet the following design requirements for standard and stretched clusters in your vSphere cluster design for VMware Cloud Foundation.The cluster design considers the storage type for the cluster, the architecture model of the environment, and the life cycle management method .

Table 4. vSphere Cluster Design Requirements for VMware Cloud Foundation

Requirement ID	Design Requirement	Justification	Implication
VCF-CLS-REQD-CFG-001	Create a cluster in each workload domain for the initial set of ESXi hosts.	Simplifies configuration by isolating management from customer workloads. Ensures that customer workloads have no impact on the management stack.	Management of multiple clusters and vCenter Server instances increases operational overhead.
VCF-CLS-REQD-CFG-002	Allocate a minimum number of ESXi hosts according to the cluster type being deployed.	Ensures correct level of redundancy to protect against host failure in the cluster.	To support redundancy, you must allocate additional ESXi host resources.
VCF-CLS-REQD-CFG-003	If using a consolidated workload domain, configure the following vSphere resource pools to control resource usage by management and customer workloads. cluster-name-rp-sddc-mgmt cluster-name-rp-sddc-edge cluster-name-rp-user-edge cluster-name-rp-user-vm	Ensures sufficient resources for the management components.	You must manage the vSphere resource pool settings over time.
VCF-CLS-REQD-CFG-004	Configure the vSAN network gateway IP address as the isolation address for the cluster.	Allows vSphere HA to validate if a host is isolated from the vSAN network.	None.
VCF-CLS-REQD-CFG-005	Set the advanced cluster setting das.usedefaultisolationaddress to false.	Ensures that vSphere HA uses the manual isolation addresses instead of the default management network gateway address.	None.

Table 5. vSphere Cluster Design Requirements for vSAN Stretched Clusters with VMware Cloud Foundation

Requirement ID	Design Requirement	Justification	Implication
VCF-CLS-REQD-CFG-006	Configure the vSAN network gateway IP addresses for the second availability zone as an additional isolation addresses for the cluster.	Allows vSphere HA to validate if a host is isolated from the vSAN network for hosts in both availability zones.	None.
VCF-CLS-REQD-CFG-007	Enable the Override default gateway for this adapter setting on the vSAN VMkernel adapters on all ESXi hosts.	Enables routing the vSAN data traffic through the vSAN network gateway rather than through the management gateway.	vSAN networks across availability zones must have a route to each other.
VCF-CLS-REQD-CFG-008	Create a host group for each availability zone and add the ESXi hosts in the zone to the respective group.	Makes it easier to manage which virtual machines run in which availability zone.	You must create and maintain VM-Host DRS group rules.

vSphere Cluster Design Recommendations for VMware Cloud Foundation

In your vSphere cluster design, you can apply certain best practices for standard and stretched clusters .

Table 6. vSphere Cluster Design Recommendations for VMware Cloud Foundation

Recommendation ID	Design Recommendation	Justification	Implication
VCF-CLS-RCMD-CFG-001	Use vSphere HA to protect all virtual machines against failures.	vSphere HA supports a robust level of protection for both ESXi host and virtual machine availability.	You must provide sufficient resources on the remaining hosts so that virtual machines can be restarted on those hosts in the event of a host outage.
VCF-CLS-RCMD-CFG-002	Set host isolation response to Power Off and restart VMs in vSphere HA.	vSAN requires that the host isolation response be set to Power Off and to restart virtual machines on available ESXi hosts.	If a false positive event occurs, virtual machines are powered off and an ESXi host is declared isolated incorrectly.
VCF-CLS-RCMD-CFG-003	Configure admission control for 1 ESXi host failure and percentage-based failover capacity.	Using the percentage-based reservation works well in situations where virtual machines have varying and sometimes significant CPU or memory reservations. vSphere automatically calculates the reserved percentage according to the number of ESXi host failures to tolerate and the number of ESXi hosts in the cluster.	In a cluster of 4 ESXi hosts, the resources of only 3 ESXi hosts are available for use.
VCF-CLS-RCMD-CFG-004	Enable VM Monitoring for each cluster.	VM Monitoring provides in-guest protection for most VM workloads. The application or service running on the virtual machine must be capable of restarting successfully after a reboot or the virtual machine restart is not sufficient.	None.
VCF-CLS-RCMD-CFG-005	Set the advanced cluster setting das.iostatsinterval to 0 to deactivate monitoring the storage and network I/O activities of the management appliances.	Enables triggering a restart of a management appliance when an OS failure occurs and heartbeats are not received from VMware Tools instead of waiting additionally for the I/O check to complete.	If you want to specifically enable I/O monitoring, you must configure the das.iostatsinterval advanced setting.
VCF-CLS-RCMD-CFG-006	Enable vSphere DRS on all clusters, using the default fully automated mode with medium threshold.	Provides the best trade-off between load balancing and unnecessary migrations with vSphere vMotion.	If a vCenter Server outage occurs, the mapping from virtual machines to ESXi hosts might be difficult to determine.
VCF-CLS-RCMD-CFG-007	Enable Enhanced vMotion Compatibility (EVC) on all clusters in the management domain.	Supports cluster upgrades without virtual machine downtime.	You must enable EVC only if the clusters contain hosts with CPUs from the same vendor. You must enable EVC on the default management domain cluster during bringup.
VCF-CLS-RCMD-CFG-008	Set the cluster EVC mode to the highest available baseline that is supported for the lowest CPU architecture on the hosts in the cluster.	Supports cluster upgrades without virtual machine downtime.	None.
VCF-CLS-RCMD-LCM-001	Use images as the life cycle management method for VI workload domains.	vSphere Lifecycle Manager images simplify the management of firmware and vendor add-ons manually.	An initial cluster image is required during workload domain or cluster deployment.

Table 7. vSphere Cluster Design Recommendations for vSAN Stretched Clusters with VMware Cloud Foundation

Recommendation ID	Design Recommendation	Justification	Implication
VCF-CLS-RCMD-CFG-009	Increase admission control percentage to half of the ESXi hosts in the cluster.	Allocating only half of a stretched cluster ensures that all VMs have enough resources if an availability zone outage occurs.	In a cluster of 8 ESXi hosts, the resources of only 4 ESXi hosts are available for use. If you add more ESXi hosts to the default management cluster, add them in pairs, one per availability zone.
VCF-CLS-RCMD-CFG-010	Create a virtual machine group for each availability zone and add the VMs in the zone to the respective group.	Ensures that virtual machines are located only in the assigned availability zone to avoid unnecessary vSphere vMotion migrations.	You must add virtual machines to the allocated group manually.
VCF-CLS-RCMD-CFG-011	Create a should-run-on-hosts-in-group VM-Host affinity rule to run each group of virtual machines on the respective group of hosts in the same availability zone.	Ensures that virtual machines are located only in the assigned availability zone to avoid unnecessary vSphere vMotion migrations.	You must manually create the rules.