VMware ESXi and vSphere Cluster Management

Create a Cluster in VMware vSphere

Learn how to create a vSphere cluster in vCenter Server, select HA, DRS, EVC, and vSAN services, add ESXi hosts, and validate the finished cluster.

What a vSphere Cluster Is

A vSphere cluster is a logical collection of ESXi hosts managed by vCenter Server as one compute resource. An ESXi host is a physical server running the VMware hypervisor and hosting virtual machines.

Each member host contributes CPU and memory capacity to the cluster's shared compute pool. A resource pool can then divide or reserve part of that capacity for particular workloads, teams, or applications.

A virtual machine still runs on one specific ESXi host at any given moment. Cluster services can decide where it should run, restart it on another host after a failure, or migrate it between hosts. vMotion performs live migration of a running virtual machine between compatible hosts.

Why Use a Cluster?

Centralized management: vCenter Server provides one inventory location for multiple ESXi hosts, their virtual machines, networks, datastores, alarms, and tasks.
Availability: vSphere HA can restart affected virtual machines on surviving hosts after an ESXi host failure.
Balancing: DRS evaluates resource demand and can recommend or automate virtual-machine placement.
Maintenance and mobility: vMotion and related compatibility features help administrators move workloads before host maintenance.
CPU compatibility: EVC can hide selected newer CPU features so supported processor generations can participate in migrations.
Shared storage: vSAN can aggregate local disks from hosts into shared datastore capacity when the complete storage architecture has been designed for it.
Continuous availability: Fault Tolerance can provide a secondary execution instance for supported virtual machines, subject to product limits and requirements.

Creating the cluster object is only the first step. HA, DRS, EVC, vSAN, and Fault Tolerance each require separate compatibility, licensing, networking, capacity, and operational decisions.

Planning and Prerequisites

Before starting, make sure the target design is understood. A cluster should be created inside a datacenter, which is a vCenter inventory container that can hold clusters, hosts, networks, storage objects, and virtual machines.

Area	Check	Why it matters
vCenter access	Confirm a vCenter Server instance, appropriate inventory permissions, and access to the target datacenter.	Clusters are vCenter inventory objects and cannot be created from an isolated ESXi host.
Host compatibility	Confirm that intended ESXi hosts are connected to the same vCenter Server and support the target vSphere version.	Version, hardware, configuration, and feature differences can prevent a host from joining.
Licensing	Verify licenses for ESXi, vCenter Server, and planned services such as HA, DRS, EVC-related functionality, vSAN, and Fault Tolerance.	A feature being visible in the client does not mean the environment is licensed to use it.
CPU and EVC planning	Compare processor generations and select an EVC baseline compatible with every current and planned host if migration across generations is required.	EVC limits exposed CPU features to a common baseline and can improve migration compatibility.
Management and VMkernel networking	Plan redundant management networking and correctly configured VMkernel networks for vMotion, storage, and vSAN where applicable.	HA, vMotion, storage access, and vSAN depend on reliable and correctly routed host networking.
Storage visibility	Confirm shared datastores or another supported storage design for workloads that will use vMotion, HA, DRS, or Fault Tolerance.	Hosts must be able to access the required virtual-machine files and dependencies.
HA spare capacity	Determine how many host failures the cluster must tolerate and reserve sufficient CPU and memory capacity.	HA admission control cannot protect workloads if the surviving hosts lack capacity.
vSAN hardware and network design	Validate supported disks, controllers, licensing, network topology, capacity, and fault domains before selecting vSAN.	vSAN is a storage architecture, not merely a cluster creation checkbox.

Review Existing Hosts and Virtual Machines

Identify virtual machines, datastores, port groups, VMkernel adapters, reservations, affinity rules, and host-specific settings before moving a standalone host into the cluster. The move can change which cluster services apply to the host and its workloads.

When required by the environment, place a host into maintenance mode before moving it or changing its configuration. Maintenance mode is a host state used to evacuate or prevent virtual-machine workloads before maintenance. Use vMotion or another approved evacuation method where possible.

Cluster Feature Selection Guide

Feature	Primary purpose	Key prerequisites	When to enable	Important caution
vSphere HA	Restart affected virtual machines on remaining hosts after a host failure.	Reliable management networking, compatible hosts, failover capacity, and appropriate datastore or heartbeat design.	Enable for workloads that require host-failure recovery.	HA does not guarantee recovery when there is insufficient spare capacity, failed management networking, or unavailable application dependencies.
DRS	Evaluate resource demand and recommend or automate virtual-machine placement.	Compatible hosts, appropriate licensing, vMotion readiness, and defined automation preferences.	Enable when balancing, placement automation, or maintenance evacuation is useful.	Reservations, affinity rules, resource pools, powered-off states, and vMotion problems can limit movement.
EVC	Present a common CPU feature baseline to support migration across supported processor differences.	A baseline supported by every cluster host and no incompatible powered-on workload requirements.	Enable before placing workloads on hosts with differing supported CPU generations.	Choose the baseline based on the oldest or least capable host; EVC does not make incompatible hardware compatible.
vSAN	Aggregate local host disks into shared software-defined storage.	Supported devices, controllers, network topology, capacity, fault domains, licensing, and a validated storage design.	Enable only when the cluster is intentionally designed to provide vSAN storage.	Do not select it as a trial option in a production cluster without a disk, network, and failure-domain plan.
vMotion	Move a running virtual machine between compatible ESXi hosts.	Reachable vMotion VMkernel networking, CPU compatibility, supported storage access, and compatible virtual hardware.	Plan it whenever workloads need mobility, DRS balancing, or non-disruptive maintenance.	A cluster does not automatically provide working vMotion; networking and workload compatibility must be tested.
Fault Tolerance	Maintain a secondary execution instance for continuous availability of supported virtual machines.	Supported product edition, hardware, networking, storage, and workload configuration.	Use only for workloads whose availability requirements justify its resource and configuration costs.	FT has workload, scale, performance, and infrastructure limitations; verify current product requirements.

Create the Cluster in the vSphere Client

Sign in to the vSphere Client connected to vCenter Server.
Open the Hosts and Clusters inventory view.
Locate and select the destination datacenter. You can also right-click the datacenter and choose the action to create a new cluster.
Start the New Cluster workflow.
Enter a meaningful, unique cluster name. Use a naming convention that identifies the site, purpose, or environment.
Choose the initial cluster service options. Select only services supported by the design, licensing, and operational plan.
Review the summary of the selected settings, then complete the wizard.
Verify that the empty or partially populated cluster object appears beneath the intended datacenter.

Depending on the vSphere version and permissions, the wizard may expose service choices in one screen or across several configuration steps. The exact labels can vary, but the decisions remain the same: name the cluster, select appropriate initial services, review, and finish.

Selecting Services After Creation

vSphere HA

Enable HA when the cluster must recover virtual machines after an ESXi host failure. Provide redundant management networking and enough unreserved capacity for the failure policy. Configure admission control, the HA capacity policy that reserves enough resources to tolerate configured failures. A three-host cluster, for example, may be designed to tolerate the loss of one host, but only if the remaining hosts have sufficient usable CPU and memory.

DRS

Enable DRS when the environment needs resource-aware placement, balancing, or automated evacuation during maintenance. Choose an automation level that matches operational requirements. Manual or recommendation-based operation gives administrators more control; more automated settings can apply approved placement actions with less intervention.

EVC

Enhanced vMotion Compatibility presents a selected CPU baseline to virtual machines. Newer hosts may support additional processor instructions, but EVC can conceal those features so workloads see a common compatible set. Select a baseline supported by every current and planned member. Powered-on virtual machines or CPU feature requirements can prevent a change, so test the setting with noncritical workloads.

vSAN

Select vSAN only after validating the storage architecture. Confirm supported disk devices and controllers, network performance and redundancy, capacity and protection requirements, licensing, and fault-domain placement. Disk claiming, storage policies, and health validation are later configuration tasks.

vMotion and Fault Tolerance

vMotion is a capability that depends on host networking, CPU compatibility, storage access, and virtual-machine compatibility. Fault Tolerance has additional product and workload limits. Treat both as design outcomes to validate rather than assuming that creating a cluster activates them automatically.

Most cluster services can generally be enabled or adjusted after creation, subject to compatibility and operational impact. Make changes through a controlled change window and review the resulting tasks, alarms, and workload effects.

Adding ESXi Hosts to the Cluster

After creating the cluster, add the intended ESXi hosts through the inventory interface. In supported vSphere Client workflows, you can drag a host onto the cluster or use the cluster's Add Hosts action. Follow the prompts to select existing hosts or connect hosts that are not yet in the required inventory state.

Review each host's connection state, vSphere version, license, hardware, CPU generation, networking, storage visibility, and current workloads.
Resolve incompatible configurations before the move. This can include port groups, VMkernel adapters, datastores, permissions, or host-specific settings.
Evacuate virtual machines or place the host into maintenance mode when required by the environment. Do not assume that moving the inventory object is safe while production workloads remain on the host.
Move or add the host to the target cluster.
Wait for vCenter tasks to complete and confirm that the host shows as connected under the cluster.
Repeat the process for each host, maintaining the intended failure-domain and capacity design.

When a host becomes a member of the cluster, its compute resources contribute to the cluster's usable capacity. Cluster services can then consider that host for placement, recovery, and migration, subject to compatibility and policy constraints.

Post-Creation Validation

Validation item	Expected result	Where to verify
Host membership	Every intended ESXi host appears under the correct cluster and no unintended host is present.	Hosts and Clusters inventory tree and cluster host list.
Host connection state	Each member is connected and reports usable CPU and memory.	Cluster Hosts view, host summary, and recent tasks.
HA status	HA is healthy, host communication is working, and admission control reflects the failure policy.	Cluster Configure or Monitor views, HA summary, capacity information, and alarms.
DRS status	DRS is enabled with the intended automation level and recommendations are understandable.	Cluster DRS configuration, recommendations, rules, and recent tasks.
EVC mode	The selected EVC baseline is active and compatible with all members.	Cluster CPU or EVC configuration and host compatibility details.
vSAN health when used	vSAN health checks pass, expected disks are claimed, and capacity and fault domains are correct.	vSAN Monitor or Health view, disk management, and capacity information.
Migration readiness	vMotion VMkernel paths, CPU compatibility, storage access, and virtual-machine requirements are valid.	Host networking, datastore views, compatibility checks, and a suitable non-disruptive test migration.
Active alarms	No unexplained cluster, host, networking, storage, or service alarms remain.	Cluster Monitor, Alarms, Tasks, and Events views.

Where the design allows, perform a controlled vMotion test with a noncritical virtual machine. Confirm that the VMkernel path is reachable, the destination host can access required storage, and the workload continues without interruption.

Document the cluster name, datacenter, member hosts, enabled features, EVC baseline, capacity assumptions, admission-control policy, storage design, networking, and failure-domain placement.

Practical Deployment Examples

Small High-Availability Compute Cluster

Create a cluster in a production datacenter, enable HA, add three similarly configured ESXi hosts, and verify that sufficient unreserved capacity exists to restart workloads after one host failure. Test management-network redundancy and document the admission-control assumption.

Balanced Virtual-Machine Placement

Enable DRS on a multi-host cluster and review its placement recommendations after adding virtual machines with uneven CPU or memory demand. If automatic actions are selected, confirm that vMotion networking and storage access are working first.

Mixed Supported CPU Generations

Before moving workloads between hosts with different supported processor generations, select an EVC baseline compatible with every cluster member. Validate migration with noncritical workloads before moving important applications.

Incremental Cluster Deployment

Create the empty cluster first and leave advanced services disabled until design validation is complete. Add hosts, verify networking and storage, then activate HA or DRS in a controlled change window.

Troubleshooting

An ESXi Host Cannot Be Added

Confirm that the host is connected to the same vCenter Server and is not already in a conflicting inventory state.
Review vSphere version, license, hardware, networking, storage, and feature compatibility.
Check whether existing workloads or host configuration require migration or maintenance mode.
Review the vCenter task and event details for the exact compatibility error.

EVC Cannot Be Enabled

Identify the oldest or least capable CPU generation among all cluster hosts.
Select a baseline supported by every current member.
Check powered-on workloads and CPU feature requirements that may block the change.

HA Shows Warnings

Verify management-network redundancy and host-to-host connectivity.
Review datastore and heartbeat configuration as applicable to the environment.
Check admission-control settings and available failover capacity.

DRS Does Not Balance Virtual Machines

Confirm that DRS is enabled and its automation level permits automatic actions if automation is expected.
Review affinity rules, reservations, resource pools, maintenance states, and other constraints.
Verify vMotion compatibility and VMkernel networking.

vMotion or DRS Migration Fails

Confirm that each host has correctly configured and reachable vMotion VMkernel networking.
Verify CPU compatibility, EVC settings, shared storage or another supported migration design, and virtual-machine device compatibility.
Review recent tasks and events for the precise migration error.

vSAN Cannot Be Enabled or Reports Health Alarms

Verify that host hardware, storage devices, networking, licensing, and the selected vSAN architecture meet requirements.
Review disk claim status, network connectivity, capacity, and fault-domain configuration.
Use vSAN health checks to identify whether the affected layer is hardware, storage, network, capacity, or configuration.

Operational Cautions

Do not enable a service solely because it is available. Align it with licensing, infrastructure design, and operational goals.
Avoid mixing incompatible CPU generations without an EVC plan.
Do not assume HA protects workloads when spare capacity is insufficient, management networking has failed, or required dependencies are unavailable.
Treat vSAN as a storage architecture decision, not a simple checkbox.
Use maintenance mode and evacuation planning before disruptive host changes.

Exam-Relevant Notes

A cluster is a vCenter-managed grouping of ESXi hosts, usually contained by a datacenter inventory object.
Virtual machines remain hosted by individual ESXi hosts even though cluster services manage placement and recovery.
HA provides restart-based recovery after host failure; it is not the same as continuous execution through Fault Tolerance.
DRS requires vMotion readiness for many balancing and maintenance actions.
EVC exposes a common CPU baseline; it does not upgrade or physically equalize host processors.
Admission control reserves capacity for the configured failure policy.
vSAN requires a complete supported storage and network design before enablement.

For the related workflow, see Create Clusters.