Architecture and Technologies
The management plane and the data plane
vSphere separates management from workload execution. ESXi is a Type-1 hypervisor that runs directly on hardware and hosts VMs; vCenter Server is the management appliance that provides inventory, cloning, DRS, and centralized permissions. A frequent exam trap is implying that VMs stop when vCenter is unavailable. They do not: ESXi hosts continue running VMs, and vSphere HA can restart VMs without vCenter because HA agents work on the hosts. Features that require vCenter, such as DRS-initiated migrations, cloning, and most inventory changes, are unavailable.
Inventory and resource hierarchy
Know the order: vCenter → datacenter → cluster → host → resource pool → VM, plus folders for organization. A cluster is the boundary for HA, DRS, EVC, and vSAN. Resource-pool shares are relative among siblings only; reservations are absolute and limits cap consumption. EVC masks CPU features so vMotion can work across different CPU generations. vSphere HA admission control reserves capacity; DRS balances load; neither replaces proper cluster sizing.
Storage and networking abstractions
VMFS is a clustered file system on block storage; NFS is file-level; vVols offloads VM operations to the array via VASA; vSAN pools local disks into a distributed datastore. Standard switches are per-host; distributed switches are managed centrally and support Network I/O Control and private VLANs.
Worked example: A cluster has 4 hosts, each with 64 GB RAM, and admission control reserves 25%. That reserves 16 GB per host for HA, leaving 48 GB usable per host. If each VM needs 8 GB, the cluster supports 24 powered-on VMs, not 32.
Rule of thumb: If the scenario asks whether VMs keep running when vCenter is down, answer yes; if it asks about DRS or cloning, answer no.