Featured image

Table of Contents Link to heading

What vCenter Server Does Link to heading

In a vSphere environment, individual ESXi hosts can be managed independently — but as the number of hosts grows, managing each one separately becomes operationally impractical. vCenter Server is the centralised management platform that aggregates all ESXi hosts and their virtual machines under a single management interface.

vCenter enables capabilities that require a central coordinator:

  • vMotion: Live migration of running VMs between ESXi hosts — requires vCenter to orchestrate the transfer
  • HA (High Availability): Automatic restart of VMs on surviving hosts when a host fails — requires vCenter to maintain cluster state
  • DRS (Distributed Resource Scheduler): Automatic load balancing across cluster hosts — requires cluster-wide visibility into resource utilisation
  • Storage vMotion: Live migration of VM storage between datastores
  • Distributed vSwitches (vDS): Network switching fabric managed centrally across all hosts
  • Unified inventory: A single view of all hosts, VMs, clusters, datastores, and networks

Without vCenter, ESXi hosts can still run VMs, but HA, DRS, vMotion, and vDS are unavailable.

Deployment Options: VCSA vs Windows Link to heading

VMware ships vCenter Server in two forms:

vCenter Server Appliance (VCSA): A pre-built Photon OS (Linux) virtual appliance. This is the current and only supported deployment option as of vSphere 7.0 — VMware deprecated the Windows-based installer.

  • Pre-configured, hardened operating system
  • Embedded PSC (Platform Services Controller) in all modern deployments
  • Built-in vPostgres database (no external database required for most deployments)
  • Simplified patching via VAMI (Virtual Appliance Management Interface)
  • Deployed from an ISO installer — the VCSA is installed onto an existing ESXi host

vCenter Server for Windows (legacy, end-of-life): Installed on a Windows Server. Required an external database (SQL Server, Oracle). Removed from vSphere 7.0+.

Note
All new vCenter deployments use VCSA. If you are managing an environment with a Windows-based vCenter, plan its migration to VCSA — VMware provides migration tools (vcsa-migrate) that transfer configuration and inventory from Windows vCenter to a new VCSA appliance.

vCenter Architecture Link to heading

vSphere Client (HTML5) Link to heading

The vSphere Client is the web-based management interface for vCenter, accessed via browser at https://vcenter-fqdn/ui. It replaced the legacy vSphere Web Client (Flash) and the deprecated Windows-based vSphere Client (C# thick client).

The HTML5 client provides:

  • VM and host management (deploy, power control, console access)
  • Cluster, network, and storage management
  • Performance charts and task history
  • vCenter configuration (users, permissions, licensing)

vCenter Server Services Link to heading

The VCSA runs several internal services:

Service Function
vCenter Server (vpxd) Core inventory, task, and event management
vSphere HA Agent Coordinates HA cluster state
DRS Load balancing decisions across cluster
vPostgres Embedded database for inventory and configuration
ESXi Dump Collector Receives VMkernel core dumps for analysis
Syslog Collector Central log aggregation from ESXi hosts
Auto Deploy PXE-based stateless ESXi host provisioning
Content Library Centralised VM template and ISO library
vCenter HA (VCHA) Control-plane redundancy for vCenter itself

Platform Services Controller (PSC) Link to heading

The PSC provides authentication and licensing services for the vSphere environment:

  • Single Sign-On (SSO): Authentication for all vSphere components — a single login authenticates to vCenter, vSAN, NSX, and other integrated products
  • VMware Certificate Authority (VMCA): Issues SSL certificates to ESXi hosts and vCenter components
  • License service: Manages vSphere, vSAN, and NSX licence keys

In vSphere 7.0+, the PSC is embedded directly into the VCSA — the separate external PSC deployment model is deprecated. One vCenter = one embedded PSC.

Data Centre and Cluster Organisation Link to heading

vCenter organises the managed environment into a logical hierarchy:

vCenter Server
└── Datacenter (logical grouping)
    ├── Cluster (group of ESXi hosts)
    │   ├── ESXi Host 1
    │   │   ├── VM-01
    │   │   └── VM-02
    │   └── ESXi Host 2
    │       └── VM-03
    ├── Standalone ESXi Host (not in cluster)
    ├── Datastore (shared storage)
    └── Network (vSS or vDS)

Datacenter: A logical container — not necessarily a physical location. It groups hosts, clusters, datastores, and networks. Access permissions can be scoped to the datacenter level.

Cluster: A group of ESXi hosts that pool their resources and can participate in HA, DRS, and vSAN. A cluster must be enabled for each feature individually.

Cluster Feature What it does Requirement
vSphere HA Restarts VMs on surviving hosts if a host fails Shared storage accessible by all hosts
DRS Automatically migrates VMs to balance CPU/memory vMotion network between hosts
vSAN Hyper-converged storage from host-local disks NVMe or SSD on each host; vSAN licence
EVC (Enhanced vMotion Compatibility) Masks CPU features to enable vMotion between different CPU generations Careful planning to avoid exposing all features
Tip
Enable EVC on new clusters before adding any VMs. Setting the EVC baseline after VMs are running requires powering them off. EVC sets the minimum CPU feature set presented to VMs, ensuring that VMs can migrate between hosts with different CPU generations within the same family (Intel or AMD).

Resource Pools Link to heading

A resource pool partitions a cluster’s (or host’s) CPU and memory into controllable pools. VMs in a resource pool share that pool’s allocation and compete only with other VMs in the same pool.

Resource pools support three control mechanisms:

Mechanism Description
Shares Relative priority — a VM with 2000 shares gets twice the resource of one with 1000 when contention occurs
Reservation Guaranteed minimum — these resources are always available to the pool, even under contention
Limit Hard cap — the pool cannot use more than this amount, even if resources are idle
Cluster (32 vCPU, 256 GB RAM)
├── Resource Pool: Production (Reservation: 20 vCPU, 200 GB RAM)
│   ├── VM: web-01 (Shares: High)
│   └── VM: db-01 (Reservation: 8 vCPU)
└── Resource Pool: Development (Limit: 4 vCPU, 32 GB RAM)
    ├── VM: dev-01
    └── VM: test-01
Warning
Overcommitting reservations is not possible — vCenter will not allow you to add a VM with a reservation that exceeds available unreserved resources in the pool. Size reservations carefully. Large reservations reduce scheduling flexibility and can leave resources idle if the reserved VMs are not using them.

vCenter Permissions and Roles Link to heading

vCenter uses a role-based access control (RBAC) model. Permissions are assigned as the combination of principal (user or group) + role (set of privileges) + inventory object (where the permission applies).

Built-in roles:

Role Typical Use
Administrator Full vCenter access
Read-Only View-only; no changes
Virtual Machine Power User Can power VMs on/off, take snapshots, configure hardware
Virtual Machine User Can use VM console and media (no configuration)
Resource Pool Administrator Manage VMs within an assigned resource pool
Datastore Consumer Allocate space on datastores
Network Administrator Manage network objects (for vDS)

Permissions propagate down the inventory hierarchy by default — a permission assigned at the datacenter level applies to all clusters, hosts, and VMs within it unless overridden at a lower level.

Best practice: assign permissions at the lowest necessary inventory level and use Active Directory groups rather than individual user accounts for manageability.

vCenter High Availability Link to heading

vCenter HA (VCHA) provides redundancy for the vCenter management plane. It creates a three-node cluster:

  • Active node: The live vCenter serving all management functions
  • Passive node: A standby replica in constant synchronisation with the active node
  • Witness node: A lightweight tie-breaker that prevents split-brain when network partitioning separates active and passive

Failover is automatic when the active node fails — the passive node takes over within minutes. Unlike ESXi HA (which restarts VMs), VCHA restarts the vCenter service itself on the passive node.

VCHA requirements:

  • All three nodes on the same vCenter (vCenter manages its own HA cluster)
  • A dedicated HA network between active and passive nodes (for replication traffic)
  • The passive node requires storage capacity equal to the active node
Note
VCHA protects against vCenter failure — it does not protect the VMs running in the managed environment. vSphere HA protects VMs from ESXi host failures. Both can (and should) be deployed simultaneously: VCHA for control-plane resilience, vSphere HA for workload resilience.

Key vCenter Operational Commands Link to heading

Access the VCSA via SSH (if enabled) or the VAMI at https://vcsa-fqdn:5480.

# Check vCenter service status
service-control --status --all

# Restart a specific service
service-control --restart vsphere-ui         # vSphere Client
service-control --restart vpxd              # vCenter Server daemon

# Check vCenter logs
tail -f /var/log/vmware/vpxd/vpxd.log
tail -f /var/log/vmware/vsphere-ui/logs/vsphere_client_virgo.log

# vCenter database health (embedded vPostgres)
/opt/vmware/vpostgres/current/bin/psql -U postgres -c "SELECT version();"

# Check VCHA status
/usr/lib/vmware-vcha/bin/vcha-status.sh

# VCSA backup via VAMI API
curl -k -u administrator@vsphere.local:password \
  -X POST "https://vcsa/api/appliance/recovery/backup/schedules"