VMware ESXi and vSphere Cluster Management

Create and Configure Clusters in VMware vSphere

Learn how to plan, create, configure, and verify a VMware vSphere cluster, enable HA and DRS, and add ESXi hosts through vCenter Server.

What a vSphere Cluster Is

A vSphere cluster is a logical vCenter Server inventory object that contains multiple ESXi hosts. An ESXi host is a physical server running the VMware hypervisor and supplying CPU, memory, networking, and storage connectivity to virtual machines.

When hosts become cluster members, their available CPU and memory capacity can be managed as a shared compute resource for virtual-machine workloads. vCenter Server manages the hosts and cluster-wide services as one logical group, while each host continues to execute virtual machines locally.

Cluster Compared with Other vSphere Objects

Object	Purpose
Datacenter	An inventory container that can contain clusters, standalone hosts, folders, networks, and datastores.
Cluster	A logical collection of ESXi hosts that provides pooled compute capacity and services such as HA and DRS.
ESXi host	A single hypervisor server that runs virtual machines and provides hardware resources.
Resource pool	A logical allocation boundary used to distribute CPU and memory from a host or cluster among workloads.
Datastore	Storage presented to ESXi for virtual-machine files and related data. A datastore is not a compute container.
Folder	An organizational inventory container. A folder does not pool compute resources or provide cluster services.

Cluster Capabilities

Cluster services solve different availability, placement, migration, compatibility, and storage problems. Enabling a service does not remove its infrastructure, compatibility, licensing, networking, or capacity requirements.

Feature	Primary purpose	Core dependencies	Key planning consideration
HA	Detects host failures and restarts affected virtual machines on surviving hosts.	Multiple hosts, usable capacity, compatible storage and networking, and appropriate licensing.	Reserve enough capacity for the expected host-failure scenario through admission control.
DRS	Evaluates demand and recommends or performs virtual-machine placement and vMotion migrations.	DRS licensing, compatible hosts, and working vMotion prerequisites for migrations.	Choose automation carefully and review rules, reservations, and migration activity.
vMotion	Live migration of a running virtual machine between compatible hosts.	vMotion VMkernel networking, CPU compatibility, datastore access where required, and compatible VM configuration.	Validate the complete migration path before relying on it for maintenance or balancing.
Fault Tolerance	Provides continuous availability for supported workloads by maintaining a secondary execution instance.	Supported VM configuration, extra compute resources, low-latency networking, storage, and suitable licensing.	FT has stricter design and resource requirements than ordinary HA restarts.
EVC	Masks selected CPU features so hosts with compatible processor generations can support migration.	A baseline supported by every intended host and compatible CPU generations.	Select the highest useful baseline that all hosts support; newer CPU features may be hidden.
vSAN	Aggregates local disks from hosts into distributed shared datastore capacity.	Supported devices, disk-group design, vSAN networking, licensing, and supported host configuration.	Enable only after validating storage, network, capacity, and fault-domain design.

Plan Before Creating the Cluster

Cluster creation is quick, but unsuitable host configuration can cause migration failures, alarms, or service disruption. Complete the following checks first.

Area	Validation item	Why it matters
vCenter inventory	Confirm the vCenter Server instance, target datacenter, and ESXi hosts are available.	The cluster must be created under a datacenter and managed by vCenter Server.
Host compatibility	Compare hardware, ESXi versions, CPU generations, firmware, and supported features.	Incompatible hosts can prevent membership, EVC, or vMotion.
Licensing	Verify feature and edition availability for HA, DRS, vMotion, FT, EVC, and vSAN.	Some services require specific licensing or supported configurations.
DNS and time	Provide consistent forward and reverse name resolution and time synchronization.	Management and cluster services depend on reliable identity and time.
Management networking	Use consistent management connectivity, VLANs, MTU settings, and administrative access.	Hosts must communicate reliably with vCenter Server and one another.
vMotion networking	Plan a VMkernel adapter and network path for vMotion on every intended host.	Live migration cannot work without a correctly configured migration network.
Shared storage or vSAN design	Verify consistent datastore presentation, or validate supported vSAN devices and network design.	Hosts need common storage access for many migration and restart scenarios.
HA capacity	Plan admission control and spare CPU and memory for the expected host failure.	Normal-state utilization alone does not prove that HA can restart workloads.
EVC requirements	Determine whether mixed processor generations require an EVC baseline.	EVC can improve compatibility but may hide newer CPU features.

Also define the cluster name, purpose, host count, expected workloads, growth capacity, maintenance windows, and DRS automation level. Review production hosts before moving them: record running virtual machines, local-only datastores, network configuration, distributed-switch membership, passthrough devices, and storage access.

Create a Cluster in the vSphere Client

Open the vSphere Client and go to Hosts and Clusters.
Locate and select the target datacenter. You can also right-click the datacenter and start the New Cluster workflow.
Enter a clear, unique cluster name that follows the organization naming standard.
Choose the initial cluster services offered by your vSphere version, such as DRS, HA, EVC, and vSAN.
Complete the wizard. The new cluster should appear beneath the intended datacenter.
Select the cluster and verify its summary, service status, and initial alarms before adding production hosts.

The exact wizard labels vary between vSphere releases. Treat the wizard as the initial configuration; review detailed policies after the object is created.

Choose Initial Cluster Settings

High Availability

Enable HA during creation when the cluster is intended to restart affected virtual machines after an ESXi host failure. After creation, review host monitoring, failure-response behavior, datastore heartbeating, isolation response, and admission control policies. Admission control reserves capacity for configured host-failure tolerance; it does not create additional capacity.

Distributed Resource Scheduler

DRS can generally operate in manual, partially automated, or fully automated modes, depending on the vSphere version and license. Manual mode produces recommendations for administrator approval. Partially automated mode can place newly powered-on virtual machines while leaving later balancing migrations for approval. Fully automated mode can perform permitted migrations automatically.

Begin with a mode appropriate for your change-control process. Check migration thresholds, affinity and anti-affinity rules, VM overrides, reservations, and workloads that should not move. Related planning may include DRS affinity rules and DRS anti-affinity rules.

Enhanced vMotion Compatibility

EVC establishes a CPU compatibility baseline by masking selected processor features exposed to virtual machines. Choose a mode supported by every host that will join the migration domain. EVC can allow migrations across compatible CPU generations, but it cannot overcome fundamentally incompatible processors or other vMotion requirements.

vSAN

Enable vSAN only after validating supported storage devices, disk groups, network segmentation, bandwidth, fault domains, capacity, and licensing. vSAN changes the storage design of the cluster; it is not a substitute for checking host networking or device compatibility.

Most cluster settings can be revised later, but changes may require workload migration, compatibility remediation, resynchronization, or service disruption. Plan changes rather than treating them as risk-free switches.

Add ESXi Hosts to the Cluster

Review each existing host and its workloads, storage, networking, version, license, and current inventory location.
In Hosts and Clusters, move the host into the target cluster by dragging it onto the cluster or using a Move To action, depending on the client version.
Accept any confirmation prompts and resolve compatibility checks before proceeding.
Add hosts one at a time, or follow a controlled migration plan for production environments.
Confirm that each host appears as a child member of the cluster and reports Connected.
Check alarms, datastore visibility, network configuration, and service compatibility after each move.

After membership is established, the host's eligible CPU and memory capacity becomes available to the cluster. This does not automatically convert local storage into shared storage or repair inconsistent VMkernel, VLAN, datastore, or security configuration.

Considerations for Existing Hosts

Running virtual machines: Decide whether workloads must be powered off, migrated, protected, or left temporarily on the host.
Local storage: Virtual machines on local-only datastores may not be movable to another host unless storage migration or another supported design is available.
Other clusters: A host must be removed from its current cluster or inventory location according to vCenter rules before joining another cluster.
Network configuration: Check standard or distributed switch membership, port groups, VMkernel adapters, VLANs, and MTU consistency.
Storage configuration: Confirm that every intended host sees required shared datastores consistently and with appropriate paths.
Special devices: Passthrough, direct-attached devices, affinity rules, and hardware-specific VM settings can prevent live migration.

Post-Creation Verification

Review the cluster summary for member-host count, total CPU and memory capacity, enabled services, alarms, and overall health.
Confirm every host is connected and can access the required datastores and networks.
Validate vMotion compatibility, including VMkernel reachability, CPU compatibility, storage access, and VM configuration.
Review HA host monitoring, admission control, datastore heartbeating, and isolation response for the environment's failure policy.
Review DRS recommendations or automation activity. Confirm that migrations are expected, permitted, and consistent with affinity rules.
Test planned functionality carefully. For example, migrate a noncritical virtual machine or place a host into maintenance mode during an approved window.

For deeper HA policy work, see admission control concepts and advanced vSphere HA options.

Practical Examples

Small High-Availability Compute Cluster

Create a cluster for three similarly configured ESXi hosts. Enable HA and DRS, connect every host to the same production datastore, and configure a vMotion VMkernel network on each host. Reserve enough CPU and memory to restart the workloads of one failed host. Test migration with a noncritical virtual machine.

Mixed CPU-Generation Hosts

Before adding older and newer compatible processor generations to one migration domain, compare CPU compatibility and select an EVC baseline supported by every host. Apply the baseline, add hosts according to the change plan, and confirm that representative virtual machines can migrate.

Adding a Standalone Host

Inspect the host's running virtual machines, local-only datastores, distributed-switch membership, and storage access. Migrate or protect workloads as required, move the host into the target cluster, and verify connected status and shared-resource access.

Preparing for Hardware Maintenance

Confirm that the cluster has sufficient capacity for the host's workloads.
Place the host in maintenance mode and migrate eligible workloads to remaining hosts.
Perform the hardware maintenance.
Reconnect the host, exit maintenance mode, and verify HA and DRS health.

Maintenance mode may not evacuate virtual machines that depend on local storage, passthrough devices, unavailable vMotion, insufficient destination capacity, or restrictive placement rules.

Operational Practices

Keep host networking, storage access, ESXi versions, patches, firmware, and security policies as uniform as practical.
Monitor capacity, resource contention, HA failover capacity, datastore health, and DRS balance over time.
Design for at least the expected host failure, not merely normal-state CPU and memory consumption.
Document enabled services, EVC mode, VMkernel networks, datastore dependencies, capacity assumptions, and the host-addition procedure.
Use maintenance mode before servicing a host and evacuate workloads according to the storage and migration capabilities of the environment.

Troubleshooting Cluster Problems

A Host Cannot Be Added

Confirm the host is connected to the same vCenter Server and is not still constrained by another cluster configuration.
Compare ESXi versions, CPU support, licenses, required networks, storage access, and hardware compatibility.
Review EVC compatibility when processor generations differ.

vMotion Is Unavailable

Verify that each host has a correctly configured VMkernel adapter with vMotion enabled.
Check IP reachability, VLANs, MTU consistency, and name-resolution requirements.
Confirm CPU compatibility, shared datastore access where required, and virtual-machine configuration compatibility.

HA Reports Insufficient Failover Resources

Review the admission-control policy and configured number of tolerated host failures.
Check CPU and memory reservations, current host capacity, and disconnected or maintenance-mode hosts.
Reduce overcommitment only after evaluating workload requirements, or add capacity.

DRS Does Not Migrate Workloads

Check whether DRS is manual, partially automated, or fully automated.
Review recommendations, migration thresholds, affinity rules, VM overrides, and host maintenance states.
Confirm that vMotion prerequisites are working.

A Host Cannot Enter Maintenance Mode

Confirm that destination hosts have enough CPU and memory.
Validate vMotion networking and datastore access.
Check anti-affinity rules, passthrough devices, local-storage dependencies, and virtual machines that cannot migrate live.

Exam-Relevant Notes

HA restarts virtual machines after a host failure; it does not provide continuous execution of the running VM.
FT provides continuous availability for supported workloads through a secondary execution instance and has additional resource and network requirements.
DRS uses vMotion for live balancing or placement migrations; DRS itself is not the migration transport.
EVC masks CPU features to improve compatibility; it does not upgrade processors or make unrelated CPUs compatible.
Admission control protects failover capacity and can prevent powering on workloads when the configured failure tolerance cannot be maintained.
A cluster is a compute-management object, whereas datastores and networks remain separate inventory and infrastructure dependencies.

Summary

To create a useful vSphere cluster, first validate vCenter inventory, host compatibility, licensing, networking, storage, and failure capacity. Create the cluster under the correct datacenter, select only services supported by the design, add ESXi hosts through the inventory interface, and verify connectivity, capacity, alarms, storage, networking, HA, DRS, and vMotion. Maintain consistent host configuration and document the design so the cluster remains reliable as workloads and hosts change.

For prerequisite learning, see the ESXi host inventory procedure, vCenter Server and ESXi communication, and VMware ESXi course topics.