HCIP-Cloud Computing key terms

A software-defined computer that runs its own operating system on emulated or virtualized hardware, isolated from other VMs sharing the same physical host.

The software layer that creates, schedules, and isolates virtual machines by allocating physical CPU, memory, storage, and network resources among them. It runs either directly on hardware or atop a host OS.

The operating system installed inside a virtual machine. It sees virtualized hardware rather than physical devices and depends on virtual drivers supplied by the hypervisor or paravirtualization interface.

A virtual processor presented to a VM. The hypervisor time-slices physical cores to serve it, so the number of vCPUs can exceed available cores, with performance governed by scheduling and overcommit ratios.

A ratio setting that lets the total vCPUs assigned to powered-on VMs exceed the host's physical logical cores. It raises consolidation density but risks contention and latency when workloads peak together.

Allocating more virtual RAM across VMs than the host physically holds, relying on techniques such as ballooning, page sharing, and swapping to reclaim memory when pressure rises.

A reclaim mechanism in which a driver inside the guest inflates a reserved balloon, forcing the guest OS to page out idle memory so the hypervisor can give those pages to other VMs.

Non-Uniform Memory Access, a server architecture where each CPU socket has local memory with faster access than remote memory. VM placement and vCPU binding should respect NUMA boundaries to avoid performance loss.

Binding a VM's vCPUs to specific physical cores so the scheduler does not migrate them. It improves determinism for latency-sensitive workloads but reduces scheduling flexibility across the cluster.

A point-in-time record of a VM's disk and, optionally, memory state that enables rollback after a failed change. Snapshots are not backups and degrade performance if retained in long chains.

A full or linked copy of an existing VM. A full clone duplicates all disk data and is independent; a linked clone shares a parent disk image and stores only differences, saving space at the cost of dependency.

A preconfigured, non-running VM image used to deploy new VMs consistently. Deploying from a template typically triggers guest customization such as hostname, IP address, and credentials.

The automated step that personalizes a VM deployed from a template or clone, setting hostname, network configuration, licensing, and domain membership so duplicates do not conflict on the network.

Moving a running VM between hosts with little or no downtime by iteratively copying memory pages while the VM keeps executing, then briefly pausing it for the final switchover.

The live migration capability in VMware environments that relocates a powered-on VM's compute state to another host over the network without disrupting service, often combined with storage migration.

Distributed Resource Scheduler, a VMware feature that continuously balances VM placement across a cluster using live migration, guided by host load, affinity rules, and configurable automation levels.

A cluster capability that detects host or VM failure and restarts affected VMs on surviving hosts. It reduces downtime but does not preserve in-memory state, so applications must be crash-consistent.

A mechanism that runs a secondary copy of a VM in lockstep with the primary so execution continues instantly if the primary host fails, trading substantial overhead for near-zero interruption.

A group of hosts managed as a single pool of compute and memory resources, enabling HA, load balancing, and shared policy enforcement across member servers.

A placement policy that keeps specified VMs together on the same host or forces them apart. Anti-affinity is commonly used to spread replicas of a tier across failure domains.

A rule requiring selected VMs to run on different hosts, protecting availability by ensuring that a single host failure cannot take down all instances of a critical service.

A logical partition of a cluster's CPU and memory that sets shares, reservations, and limits for the VMs inside it, enabling hierarchical and predictable allocation among teams or workloads.

A guaranteed minimum amount of a resource, such as CPU MHz or memory, that a VM or pool can count on. Reservations are enforced by admission control before a VM powers on.

A hard ceiling on the CPU or memory a VM may consume, used to prevent one workload from monopolizing host capacity even when spare resources exist.

A relative priority weight used to divide contended resources among VMs or pools. Shares only matter under contention and are not absolute quantities.

Allocating storage or virtual disk capacity on demand rather than reserving it upfront. It improves utilization but requires monitoring, since oversubscription can exhaust the datastore.

A logical storage container presented to hosts and VMs, backed by a VMFS or NFS volume or a local disk. It holds virtual disk files, snapshots, and configuration files.

Storage accessible to multiple hosts simultaneously, which is required for live migration and HA restarts because VM files must remain reachable from any host in the cluster.

A file or logical volume that appears to the guest as a physical disk. Its format may be thick or thin provisioned, and it can be moved or snapshotted independently of the VM configuration.

A software layer inside the host that connects VM virtual NICs to physical uplinks, providing VLAN tagging, traffic shaping, and isolation without dedicated hardware switches.

A configuration object on a virtual switch that applies consistent network policy, such as VLAN ID, security settings, and teaming, to the virtual NICs attached to it.

A Layer 2 segmentation method that isolates broadcast domains. In virtualization, VLAN IDs are tagged on port groups or virtual switch uplinks to separate management, storage, and tenant traffic.

An emulated network adapter attached to a VM and connected to a port group. Its adapter type determines driver support and performance characteristics inside the guest.

A packaged, bootable copy of an operating system and preinstalled software used to create VMs or cloud instances. Images may be stored in a library or image service and versioned.

The ability to scale compute resources up or down automatically in response to demand, adding capacity during peaks and releasing it afterward so cost tracks actual usage.

Adding or removing instances of a service to handle load, as opposed to enlarging a single machine. It suits stateless tiers and pairs naturally with load balancing.

Increasing or decreasing the CPU, memory, or storage assigned to a single instance. It is simple but usually requires a restart and is bounded by the largest available flavor.

A policy-driven service that adds or removes instances based on metrics such as CPU utilization or queue depth, maintaining performance while avoiding idle capacity.

A component that distributes incoming requests across multiple backend instances using algorithms such as round robin or least connections, and removes unhealthy members from rotation.

A periodic probe that tests whether an instance or endpoint is functioning, so a load balancer or orchestrator can stop sending traffic to failed members and restore it after recovery.

A predefined template specifying the vCPU, memory, and sometimes disk or accelerator resources for an instance. Choosing a flavor determines the compute capacity a VM receives.