Nutanix NCP-MCI cheat sheet

Prism Element runs on every CVM and manages one cluster; Prism Central is a separate VM (or scale-out set of VMs) that manages many clusters via registered Prism Element instances.

Foundation (the imaging tool) discovers nodes over IPMI/Redfish out-of-band, wipes disks, installs the hypervisor and AOS, and creates the cluster; it runs from a laptop or jump host, not from a cluster node.

A cluster requires a minimum of 3 nodes for RF2 (or 4 nodes for RF3); single-node clusters exist but offer no redundancy and are limited to specific use cases.

The default CVM memory is 16 GB (32 GB for large-node or specific workloads); the CVM reserves memory from the host, so size the host for guest VMs plus CVM overhead.

Cluster creation requires: cluster name, virtual IP (VIP), CVM IPs, hypervisor IPs, subnet mask, gateway, DNS, NTP, and an Active Directory (optional) configuration.

The cluster VIP must be a free IP in the same subnet as the CVMs; the CVM and hypervisor management addresses must be unique and outside any DHCP pool used by guest VMs.

AOS and hypervisor version compatibility is enforced by the compatibility matrix; mismatched versions block upgrades and can block cluster expansion.

After Foundation, run NCC (Nutanix Cluster Check) health checks and verify the cluster is in a healthy state before adding workloads; a new cluster starts with a default storage pool and container.

Prism Element dashboards show cluster CPU, memory, storage, and IOPS; Prism Central adds multi-cluster dashboards, categories, and policies (security, protection, recovery).

NCC (Nutanix Cluster Check) runs a library of health checks; schedule it and review failures - a failed check does not automatically mean an outage, but it should be triaged.

Alerts have severity levels (Critical, Warning, Info); alerts can be routed to email, SNMP, syslog, or webhooks, and can be acknowledged or resolved in Prism.

The health dashboard surfaces checks such as CVM memory pressure, disk health, and time synchronization; NTP drift is a common cause of cluster-wide issues.

Pulse (formerly CallHome) sends telemetry to Nutanix support; it can be disabled, but disabling it limits proactive support and automated health insights.

LCM (Life Cycle Manager) handles AOS, hypervisor, firmware, and NCC updates; it performs pre-checks, staged upgrades, and can roll back in some cases.

Cluster-wide capacity is reported as usable vs. raw; RF2 consumes roughly 2x the data footprint (plus overhead), so plan capacity with the replication factor in mind.

Prism Central registration requires network reachability from the PC to each PE VIP; a cluster can be registered to only one Prism Central at a time.

The default bridge (br0) on each host carries CVM and hypervisor management traffic; guest VM traffic uses a separate bridge or VLAN-backed network.

Nutanix AHV networks are created in Prism as VLAN-backed virtual networks; a VM NIC is attached to a network, and the VLAN ID determines the broadcast domain.

Bond modes: active-backup (default, no switch config), balance-slb (no LACP, uses source MAC hashing), and balance-tcp (requires LACP on the switch).

balance-tcp requires a properly configured LACP port-channel on the upstream switch; misconfiguration causes connectivity loss and is a common deployment error.

The cluster VIP and CVM IPs must be on a management VLAN; guest VM networks should be isolated from management traffic using separate VLANs.

AHV supports Open vSwitch (OVS) for network virtualization; flow policies and microsegmentation are configured via Prism Central (Flow).

Check network health with the network visualization in Prism, ping/arp tests from the CVM, and the 'network' NCC checks; a failed CVM-to-CVM ping indicates a bridge or VLAN issue.

For ESXi clusters, the CVM uses a vSwitch port group; the recommended configuration is a dedicated vSwitch or port group with promiscuous mode and forged transmits enabled.

A storage pool is a grouping of physical disks (HDD/SSD/NVMe); a container is a logical namespace (datastore) presented to the hypervisor, and multiple containers can share a storage pool.

Replication factor (RF) 2 tolerates one node/disk failure; RF3 tolerates two. RF is set per container (or per cluster default) and consumes capacity accordingly.

Erasure coding (EC-X) reduces capacity overhead for cold data; it is configured per container and requires a minimum number of nodes (typically 4+).

The distributed storage fabric (DSF) uses a log-structured, extent-based layout; data is written to the local CVM's staging area, then replicated to peers before acknowledgment.

Deduplication (capacity and/or inline) and compression are container-level settings; compression is inline by default on supported AOS versions, dedup is often post-process or capacity-tier.

Snapshots are metadata-based and space-efficient; they are not full copies, but changed blocks consume space over time.

Storage QoS (IOPS limits) can be applied per VM or per container to prevent noisy-neighbor effects.

Check storage health via Prism storage dashboard, NCC disk checks, and the 'ncli' commands; a disk in a 'bad' state triggers data rebuild, which consumes cluster resources.

In AHV, VMs are created from images (ISOs, disk images) stored in a container; images can be uploaded via Prism or the 'ncli' / API.

VM categories (key-value pairs) drive policy application in Prism Central for protection, recovery, and security; categories are the recommended way to group VMs.

AHV supports live migration (with shared storage or with the Acropolis Dynamic Scheduler), HA (restart on host failure), and affinity/anti-affinity rules.

VM affinity rules keep VMs on the same host; anti-affinity rules spread them across hosts. Rules are enforced by the scheduler and can be 'should' or 'must' strength.

Prism Central Projects provide self-service VM provisioning with quotas, role-based access, and environment isolation.

Guest tools (Nutanix Guest Tools, NGT) enable file-level restore, application-consistent snapshots, and self-service restore for supported guest OSes.

VM power operations, console access, and snapshot/clone actions are available in Prism Element; bulk operations and policies are in Prism Central.

A VM clone can be full or linked (dependent on the parent snapshot); linked clones save space but depend on the parent image/snapshot chain.

Protection domains group VMs (by category) for replication and snapshot policies; a protection policy defines RPO, retention, and replication target.

Async replication sends snapshots to a remote cluster at the RPO interval; sync replication (Metro Availability) requires low-latency links and provides near-zero RPO.

Nutanix DR (Disaster Recovery) uses protection policies in Prism Central; a recovery plan defines the boot order and network mapping for failover.

Snapshots are local and fast; replication is remote. A protection policy can take local snapshots only, or local plus remote replication.

NearSync replication provides RPO in seconds (typically 1-5 minutes) and requires a compatible AOS version and network latency budget.

For file-level restore, NGT must be installed in the guest; self-service restore lets users recover their own files from a snapshot.

Test failover (planned) is non-disruptive; unplanned failover promotes the remote copy. Always validate the recovery plan before relying on it.

Retention policies define how many snapshots/replicas to keep (e.g., hourly/daily/weekly); retention is applied per protection policy and consumes capacity on both sites.

NCC is the first diagnostic step: run 'ncc' from the CVM CLI or via Prism; failed checks often point to the root cause (time sync, disk, network).

CVM memory pressure is a common cause of cluster instability; monitor CVM memory and resize if consistently high.

A disk failure triggers a rebuild; monitor rebuild progress in Prism and avoid maintenance operations during a rebuild.

NTP drift causes authentication, replication, and cluster services failures; verify NTP on CVMs and hypervisors with 'ntpq -p' or 'chronyc sources'.

Log collection: use 'ncli' or Prism to generate a log bundle; logs are also available on the CVM under /home/nutanix/data/logs.

LCM pre-checks must pass before upgrades; common blockers are insufficient disk space, NCC failures, and version incompatibility.

A node in maintenance mode evacuates VMs and data; use it before hardware service to avoid data loss and to keep RF satisfied.

Cluster health is not just 'green lights' - check the health dashboard, NCC, and capacity trends; a cluster can be functional but degraded (e.g., RF not satisfied).

Prism Central is the multicloud control plane; it manages multiple Prism Element clusters (on-prem and in public cloud) from one UI.

Nutanix Clusters on AWS/Azure extends AOS networking and management to public cloud; it is not the same as running AHV on-prem.

Calm (Nutanix Cloud Manager) provides orchestration and self-service for multi-cloud workloads; blueprints define multi-tier applications.

Nutanix Cloud Manager (NCM) includes cost governance, security, and self-service; it integrates with Prism Central and public cloud accounts.

Flow (microsegmentation) policies are defined in Prism Central and applied across clusters; they use categories, not IP addresses, for policy targets.

Multi-cluster management requires network connectivity from Prism Central to each PE VIP and consistent time sync; a cluster can be registered to only one PC.

Categories are the key mechanism for applying policies (protection, security, cost) consistently across clusters and clouds.

Cloud-native workloads can be managed alongside AHV VMs; Nutanix provides CSI/CNI integrations for Kubernetes (NKP) in multicloud scenarios.

Foundation discovers nodes via IPMI/Redfish out-of-band; it wipes disks and installs the hypervisor and AOS, then creates the cluster from a jump host.

Minimum 3 nodes for RF2, 4 nodes for RF3; single-node clusters exist but provide no redundancy.

Cluster creation requires cluster name, VIP, CVM IPs, hypervisor IPs, subnet mask, gateway, DNS, and NTP; the VIP must be a free IP in the CVM subnet.

Default CVM memory is 16 GB (32 GB for large nodes); the CVM reserves host memory, so size hosts for guest VMs plus CVM overhead.

AOS and hypervisor version compatibility is enforced by the compatibility matrix; mismatches block upgrades and cluster expansion.

After Foundation, run NCC health checks and confirm a healthy state before adding workloads; a new cluster starts with a default storage pool and container.

Prism Element manages one cluster; Prism Central is a separate VM (or scale-out set) that manages many clusters via registered Prism Element instances.

The cluster VIP, CVM IPs, and hypervisor management IPs must be unique and outside any DHCP pool used by guest VMs.

Prism Element shows cluster CPU, memory, storage, and IOPS for one cluster; Prism Central adds multi-cluster dashboards, categories, and policies.

NCC (Nutanix Cluster Check) runs a library of health checks; schedule it and triage failures - a failed check does not automatically mean an outage.

Alerts have severity levels (Critical, Warning, Info) and can route to email, SNMP, syslog, or webhooks; they can be acknowledged or resolved in Prism.

NTP drift is a common cause of cluster-wide issues; verify time sync across CVMs and hypervisors.

Pulse (formerly CallHome) sends telemetry to Nutanix support; disabling it limits proactive support and automated health insights.

LCM (Life Cycle Manager) handles AOS, hypervisor, firmware, and NCC updates with pre-checks and staged upgrades.

Cluster capacity is reported as usable vs. raw; RF2 consumes roughly 2x the data footprint plus overhead, so plan capacity with RF in mind.

A cluster can be registered to only one Prism Central at a time; PC must reach each PE VIP over the network.

The default bridge (br0) on each host carries CVM and hypervisor management traffic; guest VM traffic uses a separate bridge or VLAN-backed network.

AHV networks are VLAN-backed virtual networks; a VM NIC attaches to a network, and the VLAN ID defines the broadcast domain.

Bond modes: active-backup (default, no switch config), balance-slb (no LACP, source MAC hashing), and balance-tcp (requires LACP on the switch).

balance-tcp requires a correctly configured LACP port-channel upstream; misconfiguration causes connectivity loss and is a common deployment error.

The cluster VIP and CVM IPs belong on a management VLAN; isolate guest VM networks from management traffic using separate VLANs.

AHV uses Open vSwitch (OVS); Flow microsegmentation policies are defined in Prism Central using categories.

Diagnose with Prism network visualization, CVM ping/arp tests, and NCC network checks; failed CVM-to-CVM ping indicates a bridge or VLAN issue.

On ESXi clusters, the CVM uses a vSwitch port group; enable promiscuous mode and forged transmits on the dedicated port group.

A storage pool groups physical disks (HDD/SSD/NVMe); a container is a logical namespace (datastore) presented to the hypervisor, and multiple containers can share a pool.

Replication factor (RF) 2 tolerates one node/disk failure; RF3 tolerates two. RF is set per container or as a cluster default and consumes capacity accordingly.

Erasure coding (EC-X) reduces capacity overhead for cold data; it is configured per container and requires a minimum number of nodes (typically 4+).

The distributed storage fabric (DSF) is log-structured and extent-based; writes land in the local CVM staging area, then replicate to peers before acknowledgment.

Deduplication (capacity and/or inline) and compression are container-level settings; compression is inline by default on supported AOS versions.

Snapshots are metadata-based and space-efficient; they are not full copies, but changed blocks consume space over time.

Storage QoS (IOPS limits) can be applied per VM or per container to prevent noisy-neighbor effects.

Check storage health via the Prism storage dashboard, NCC disk checks, and ncli; a disk in a bad state triggers a rebuild that consumes cluster resources.

In AHV, VMs are created from images (ISOs, disk images) stored in a container; images can be uploaded via Prism, ncli, or the API.

Categories (key-value pairs) drive policy application in Prism Central for protection, recovery, and security; use categories to group VMs.

AHV supports live migration, HA (restart on host failure), and affinity/anti-affinity rules enforced by the scheduler.

VM affinity rules keep VMs on the same host; anti-affinity rules spread them across hosts. Rules can be 'should' or 'must' strength.

Prism Central Projects provide self-service VM provisioning with quotas, role-based access, and environment isolation.

Nutanix Guest Tools (NGT) enables file-level restore, application-consistent snapshots, and self-service restore for supported guest OSes.

VM power operations, console access, and snapshot/clone actions are available in Prism Element; bulk operations and policies are in Prism Central.

A VM clone can be full or linked (dependent on the parent snapshot); linked clones save space but depend on the parent image/snapshot chain.

Protection domains group VMs (by category) for replication and snapshot policies; a protection policy defines RPO, retention, and replication target.

Async replication sends snapshots to a remote cluster at the RPO interval; sync replication (Metro Availability) requires low-latency links and provides near-zero RPO.

Nutanix DR uses protection policies in Prism Central; a recovery plan defines boot order and network mapping for failover.

Snapshots are local and fast; replication is remote. A protection policy can take local snapshots only, or local plus remote replication.

NearSync replication provides RPO in seconds (typically 1-5 minutes) and requires a compatible AOS version and network latency budget.

For file-level restore, NGT must be installed in the guest; self-service restore lets users recover their own files from a snapshot.

Test failover (planned) is non-disruptive; unplanned failover promotes the remote copy. Validate the recovery plan before relying on it.

Retention policies define how many snapshots/replicas to keep (e.g., hourly/daily/weekly); retention applies per protection policy and consumes capacity on both sites.

NCC is the first diagnostic step: run it from the CVM CLI or via Prism; failed checks often point to the root cause (time sync, disk, network).

CVM memory pressure is a common cause of cluster instability; monitor CVM memory and resize if consistently high.

A disk failure triggers a rebuild; monitor rebuild progress in Prism and avoid maintenance operations during a rebuild.

NTP drift causes authentication, replication, and cluster services failures; verify NTP on CVMs and hypervisors with ntpq -p or chronyc sources.

Log collection: generate a log bundle via ncli or Prism; logs are also on the CVM under /home/nutanix/data/logs.

LCM pre-checks must pass before upgrades; common blockers are insufficient disk space, NCC failures, and version incompatibility.

A node in maintenance mode evacuates VMs and data; use it before hardware service to avoid data loss and keep RF satisfied.

Cluster health is not just green lights - check the health dashboard, NCC, and capacity trends; a cluster can be functional but degraded (e.g., RF not satisfied).

Prism Central is the multicloud control plane; it manages multiple Prism Element clusters (on-prem and in public cloud) from one UI.

Nutanix Clusters on AWS/Azure extends AOS networking and management to public cloud; it is not the same as running AHV on-prem.

Calm (Nutanix Cloud Manager) provides orchestration and self-service for multi-cloud workloads; blueprints define multi-tier applications.

Nutanix Cloud Manager (NCM) includes cost governance, security, and self-service; it integrates with Prism Central and public cloud accounts.

Flow (microsegmentation) policies are defined in Prism Central and applied across clusters; they use categories, not IP addresses, for policy targets.

Multi-cluster management requires network connectivity from Prism Central to each PE VIP and consistent time sync; a cluster can be registered to only one PC.

Categories are the key mechanism for applying policies (protection, security, cost) consistently across clusters and clouds.

Cloud-native workloads can be managed alongside AHV VMs; Nutanix provides CSI/CNI integrations for Kubernetes (NKP) in multicloud scenarios.

Foundation discovers nodes via IPMI/Redfish out-of-band; it wipes disks, installs the hypervisor and AOS, and creates the cluster from a jump host.

Minimum 3 nodes for RF2, 4 nodes for RF3; single-node clusters exist but provide no redundancy.

Cluster creation requires cluster name, VIP, CVM IPs, hypervisor IPs, subnet mask, gateway, DNS, and NTP; the VIP must be a free IP in the CVM subnet.

Default CVM memory is 16 GB (32 GB for large nodes); the CVM reserves host memory, so size hosts for guest VMs plus CVM overhead.

AOS and hypervisor version compatibility is enforced by the compatibility matrix; mismatches block upgrades and cluster expansion.

After Foundation, run NCC health checks and confirm a healthy state before adding workloads; a new cluster starts with a default storage pool and container.

Prism Element manages one cluster; Prism Central is a separate VM (or scale-out set) that manages many clusters via registered Prism Element instances.

The cluster VIP, CVM IPs, and hypervisor management IPs must be unique and outside any DHCP pool used by guest VMs.

Prism Element shows cluster CPU, memory, storage, and IOPS for one cluster; Prism Central adds multi-cluster dashboards, categories, and policies.

NCC (Nutanix Cluster Check) runs a library of health checks; schedule it and triage failures - a failed check does not automatically mean an outage.

Alerts have severity levels (Critical, Warning, Info) and can route to email, SNMP, syslog, or webhooks; they can be acknowledged or resolved in Prism.

NTP drift is a common cause of cluster-wide issues; verify time sync across CVMs and hypervisors.

Pulse (formerly CallHome) sends telemetry to Nutanix support; disabling it limits proactive support and automated health insights.

LCM (Life Cycle Manager) handles AOS, hypervisor, firmware, and NCC updates with pre-checks and staged upgrades.

Cluster capacity is reported as usable vs. raw; RF2 consumes roughly 2x the data footprint plus overhead, so plan capacity with RF in mind.

A cluster can be registered to only one Prism Central at a time; PC must reach each PE VIP over the network.

The default bridge (br0) on each host carries CVM and hypervisor management traffic; guest VM traffic uses a separate bridge or VLAN-backed network.

AHV networks are VLAN-backed virtual networks; a VM NIC attaches to a network, and the VLAN ID defines the broadcast domain.

Bond modes: active-backup (default, no switch config), balance-slb (no LACP, source MAC hashing), and balance-tcp (requires LACP on the switch).

balance-tcp requires a correctly configured LACP port-channel upstream; misconfiguration causes connectivity loss and is a common deployment error.

The cluster VIP and CVM IPs belong on a management VLAN; isolate guest VM networks from management traffic using separate VLANs.

AHV uses Open vSwitch (OVS); Flow microsegmentation policies are defined in Prism Central using categories.

Diagnose with Prism network visualization, CVM ping/arp tests, and NCC network checks; failed CVM-to-CVM ping indicates a bridge or VLAN issue.

On ESXi clusters, the CVM uses a vSwitch port group; enable promiscuous mode and forged transmits on the dedicated port group.

A storage pool groups physical disks (HDD/SSD/NVMe); a container is a logical namespace (datastore) presented to the hypervisor, and multiple containers can share a pool.

Replication factor (RF) 2 tolerates one node/disk failure; RF3 tolerates two. RF is set per container or as a cluster default and consumes capacity accordingly.

Erasure coding (EC-X) reduces capacity overhead for cold data; it is configured per container and requires a minimum number of nodes (typically 4+).

The distributed storage fabric (DSF) is log-structured and extent-based; writes land in the local CVM staging area, then replicate to peers before acknowledgment.

Deduplication (capacity and/or inline) and compression are container-level settings; compression is inline by default on supported AOS versions.

Snapshots are metadata-based and space-efficient; they are not full copies, but changed blocks consume space over time.

Storage QoS (IOPS limits) can be applied per VM or per container to prevent noisy-neighbor effects.

Check storage health via the Prism storage dashboard, NCC disk checks, and ncli; a disk in a bad state triggers a rebuild that consumes cluster resources.

In AHV, VMs are created from images (ISOs, disk images) stored in a container; images can be uploaded via Prism, ncli, or the API.

Categories (key-value pairs) drive policy application in Prism Central for protection, recovery, and security; use categories to group VMs.

AHV supports live migration, HA (restart on host failure), and affinity/anti-affinity rules enforced by the scheduler.

VM affinity rules keep VMs on the same host; anti-affinity rules spread them across hosts. Rules can be 'should' or 'must' strength.

Prism Central Projects provide self-service VM provisioning with quotas, role-based access, and environment isolation.

Nutanix Guest Tools (NGT) enables file-level restore, application-consistent snapshots, and self-service restore for supported guest OSes.

VM power operations, console access, and snapshot/clone actions are available in Prism Element; bulk operations and policies are in Prism Central.

A VM clone can be full or linked (dependent on the parent snapshot); linked clones save space but depend on the parent image/snapshot chain.

Protection domains group VMs (by category) for replication and snapshot policies; a protection policy defines RPO, retention, and replication target.

Async replication sends snapshots to a remote cluster at the RPO interval; sync replication (Metro Availability) requires low-latency links and provides near-zero RPO.

Nutanix DR uses protection policies in Prism Central; a recovery plan defines boot order and network mapping for failover.

Snapshots are local and fast; replication is remote. A protection policy can take local snapshots only, or local plus remote replication.

NearSync replication provides RPO in seconds (typically 1-5 minutes) and requires a compatible AOS version and network latency budget.

For file-level restore, NGT must be installed in the guest; self-service restore lets users recover their own files from a snapshot.

Test failover (planned) is non-disruptive; unplanned failover promotes the remote copy. Validate the recovery plan before relying on it.

Retention policies define how many snapshots/replicas to keep (e.g., hourly/daily/weekly); retention applies per protection policy and consumes capacity on both sites.

NCC is the first diagnostic step: run it from the CVM CLI or via Prism; failed checks often point to the root cause (time sync, disk, network).

CVM memory pressure is a common cause of cluster instability; monitor CVM memory and resize if consistently high.

A disk failure triggers a rebuild; monitor rebuild progress in Prism and avoid maintenance operations during a rebuild.

NTP drift causes authentication, replication, and cluster services failures; verify NTP on CVMs and hypervisors with ntpq -p or chronyc sources.

Log collection: generate a log bundle via ncli or Prism; logs are also on the CVM under /home/nutanix/data/logs.

LCM pre-checks must pass before upgrades; common blockers are insufficient disk space, NCC failures, and version incompatibility.

A node in maintenance mode evacuates VMs and data; use it before hardware service to avoid data loss and keep RF satisfied.

Cluster health is not just green lights - check the health dashboard, NCC, and capacity trends; a cluster can be functional but degraded (e.g., RF not satisfied).

Prism Central is the multicloud control plane; it manages multiple Prism Element clusters (on-prem and in public cloud) from one UI.

Nutanix Clusters on AWS/Azure extends AOS networking and management to public cloud; it is not the same as running AHV on-prem.

Calm (Nutanix Cloud Manager) provides orchestration and self-service for multi-cloud workloads; blueprints define multi-tier applications.

Nutanix Cloud Manager (NCM) includes cost governance, security, and self-service; it integrates with Prism Central and public cloud accounts.

Flow (microsegmentation) policies are defined in Prism Central and applied across clusters; they use categories, not IP addresses, for policy targets.

Multi-cluster management requires network connectivity from Prism Central to each PE VIP and consistent time sync; a cluster can be registered to only one PC.

Categories are the key mechanism for applying policies (protection, security, cost) consistently across clusters and clouds.

Cloud-native workloads can be managed alongside AHV VMs; Nutanix provides CSI/CNI integrations for Kubernetes (NKP) in multicloud scenarios.

Foundation discovers nodes via IPMI/Redfish out-of-band; it wipes disks, installs the hypervisor and AOS, and creates the cluster from a jump host.

Minimum 3 nodes for RF2, 4 nodes for RF3; single-node clusters exist but provide no redundancy.

Cluster creation requires cluster name, VIP, CVM IPs, hypervisor IPs, subnet mask, gateway, DNS, and NTP; the VIP must be a free IP in the CVM subnet.

Default CVM memory is 16 GB (32 GB for large nodes); the CVM reserves host memory, so size hosts for guest VMs plus CVM overhead.

AOS and hypervisor version compatibility is enforced by the compatibility matrix; mismatches block upgrades and cluster expansion.

After Foundation, run NCC health checks and confirm a healthy state before adding workloads; a new cluster starts with a default storage pool and container.

Prism Element manages one cluster; Prism Central is a separate VM (or scale-out set) that manages many clusters via registered Prism Element instances.

The cluster VIP, CVM IPs, and hypervisor management IPs must be unique and outside any DHCP pool used by guest VMs.

Prism Element shows cluster CPU, memory, storage, and IOPS for one cluster; Prism Central adds multi-cluster dashboards, categories, and policies.

NCC (Nutanix Cluster Check) runs a library of health checks; schedule it and triage failures - a failed check does not automatically mean an outage.

Alerts have severity levels (Critical, Warning, Info) and can route to email, SNMP, syslog, or webhooks; they can be acknowledged or resolved in Prism.

NTP drift is a common cause of cluster-wide issues; verify time sync across CVMs and hypervisors.

Pulse (formerly CallHome) sends telemetry to Nutanix support; disabling it limits proactive support and automated health insights.

LCM (Life Cycle Manager) handles AOS, hypervisor, firmware, and NCC updates with pre-checks and staged upgrades.

Cluster capacity is reported as usable vs. raw; RF2 consumes roughly 2x the data footprint plus overhead, so plan capacity with RF in mind.

A cluster can be registered to only one Prism Central at a time; PC must reach each PE VIP over the network.

The default bridge (br0) on each host carries CVM and hypervisor management traffic; guest VM traffic uses a separate bridge or VLAN-backed network.

AHV networks are VLAN-backed virtual networks; a VM NIC attaches to a network, and the VLAN ID defines the broadcast domain.

Bond modes: active-backup (default, no switch config), balance-slb (no LACP, source MAC hashing), and balance-tcp (requires LACP on the switch).

balance-tcp requires a correctly configured LACP port-channel upstream; misconfiguration causes connectivity loss and is a common deployment error.

The cluster VIP and CVM IPs belong on a management VLAN; isolate guest VM networks from management traffic using separate VLANs.

AHV uses Open vSwitch (OVS); Flow microsegmentation policies are defined in Prism Central using categories.

Diagnose with Prism network visualization, CVM ping/arp tests, and NCC network checks; failed CVM-to-CVM ping indicates a bridge or VLAN issue.

On ESXi clusters, the CVM uses a vSwitch port group; enable promiscuous mode and forged transmits on the dedicated port group.

A storage pool groups physical disks (HDD/SSD/NVMe); a container is a logical namespace (datastore) presented to the hypervisor, and multiple containers can share a pool.

Replication factor (RF) 2 tolerates one node/disk failure; RF3 tolerates two. RF is set per container or as a cluster default and consumes capacity accordingly.

Erasure coding (EC-X) reduces capacity overhead for cold data; it is configured per container and requires a minimum number of nodes (typically 4+).

The distributed storage fabric (DSF) is log-structured and extent-based; writes land in the local CVM staging area, then replicate to peers before acknowledgment.

Deduplication (capacity and/or inline) and compression are container-level settings; compression is inline by default on supported AOS versions.

Snapshots are metadata-based and space-efficient; they are not full copies, but changed blocks consume space over time.

Storage QoS (IOPS limits) can be applied per VM or per container to prevent noisy-neighbor effects.

Check storage health via the Prism storage dashboard, NCC disk checks, and ncli; a disk in a bad state triggers a rebuild that consumes cluster resources.

In AHV, VMs are created from images (ISOs, disk images) stored in a container; images can be uploaded via Prism, ncli, or the API.

Categories (key-value pairs) drive policy application in Prism Central for protection, recovery, and security; use categories to group VMs.

AHV supports live migration, HA (restart on host failure), and affinity/anti-affinity rules enforced by the scheduler.

VM affinity rules keep VMs on the same host; anti-affinity rules spread them across hosts. Rules can be 'should' or 'must' strength.

Prism Central Projects provide self-service VM provisioning with quotas, role-based access, and environment isolation.

Nutanix Guest Tools (NGT) enables file-level restore, application-consistent snapshots, and self-service restore for supported guest OSes.

VM power operations, console access, and snapshot/clone actions are available in Prism Element; bulk operations and policies are in Prism Central.

A VM clone can be full or linked (dependent on the parent snapshot); linked clones save space but depend on the parent image/snapshot chain.

Protection domains group VMs (by category) for replication and snapshot policies; a protection policy defines RPO, retention, and replication target.

Async replication sends snapshots to a remote cluster at the RPO interval; sync replication (Metro Availability) requires low-latency links and provides near-zero RPO.

Nutanix DR uses protection policies in Prism Central; a recovery plan defines boot order and network mapping for failover.

Snapshots are local and fast; replication is remote. A protection policy can take local snapshots only, or local plus remote replication.

NearSync replication provides RPO in seconds (typically 1-5 minutes) and requires a compatible AOS version and network latency budget.

For file-level restore, NGT must be installed in the guest; self-service restore lets users recover their own files from a snapshot.

Test failover (planned) is non-disruptive; unplanned failover promotes the remote copy. Validate the recovery plan before relying on it.

Retention policies define how many snapshots/replicas to keep (e.g., hourly/daily/weekly); retention applies per protection policy and consumes capacity on both sites.

NCC is the first diagnostic step: run it from the CVM CLI or via Prism; failed checks often point to the root cause (time sync, disk, network).

CVM memory pressure is a common cause of cluster instability; monitor CVM memory and resize if consistently high.

A disk failure triggers a rebuild; monitor rebuild progress in Prism and avoid maintenance operations during a rebuild.

NTP drift causes authentication, replication, and cluster services failures; verify NTP on CVMs and hypervisors with ntpq -p or chronyc sources.

Log collection: generate a log bundle via ncli or Prism; logs are also on the CVM under /home/nutanix/data/logs.

LCM pre-checks must pass before upgrades; common blockers are insufficient disk space, NCC failures, and version incompatibility.

A node in maintenance mode evacuates VMs and data; use it before hardware service to avoid data loss and keep RF satisfied.

Cluster health is not just green lights - check the health dashboard, NCC, and capacity trends; a cluster can be functional but degraded (e.g., RF not satisfied).

Prism Central is the multicloud control plane; it manages multiple Prism Element clusters (on-prem and in public cloud) from one UI.

Nutanix Clusters on AWS/Azure extends AOS networking and management to public cloud; it is not the same as running AHV on-prem.

Calm (Nutanix Cloud Manager) provides orchestration and self-service for multi-cloud workloads; blueprints define multi-tier applications.

Nutanix Cloud Manager (NCM) includes cost governance, security, and self-service; it integrates with Prism Central and public cloud accounts.

Flow (microsegmentation) policies are defined in Prism Central and applied across clusters; they use categories, not IP addresses, for policy targets.

Multi-cluster management requires network connectivity from Prism Central to each PE VIP and consistent time sync; a cluster can be registered to only one PC.

Categories are the key mechanism for applying policies (protection, security, cost) consistently across clusters and clouds.

Cloud-native workloads can be managed alongside AHV VMs; Nutanix provides CSI/CNI integrations for Kubernetes (NKP) in multicloud scenarios.

Foundation discovers nodes via IPMI/Redfish out-of-band; it wipes disks, installs the hypervisor and AOS, and creates the cluster from a jump host.

Minimum 3 nodes for RF2, 4 nodes for RF3; single-node clusters exist but provide no redundancy.

Cluster creation requires cluster name, VIP, CVM IPs, hypervisor IPs, subnet mask, gateway, DNS, and NTP; the VIP must be a free IP in the CVM subnet.

Default CVM memory is 16 GB (32 GB for large nodes); the CVM reserves host memory, so size hosts for guest VMs plus CVM overhead.

AOS and hypervisor version compatibility is enforced by the compatibility matrix; mismatches block upgrades and cluster expansion.

After Foundation, run NCC health checks and confirm a healthy state before adding workloads; a new cluster starts with a default storage pool and container.

Prism Element manages one cluster; Prism Central is a separate VM (or scale-out set) that manages many clusters via registered Prism Element instances.

The cluster VIP, CVM IPs, and hypervisor management IPs must be unique and outside any DHCP pool used by guest VMs.

Prism Element shows cluster CPU, memory, storage, and IOPS for one cluster; Prism Central adds multi-cluster dashboards, categories, and policies.

NCC (Nutanix Cluster Check) runs a library of health checks; schedule it and triage failures - a failed check does not automatically mean an outage.

Alerts have severity levels (Critical, Warning, Info) and can route to email, SNMP, syslog, or webhooks; they can be acknowledged or resolved in Prism.

NTP drift is a common cause of cluster-wide issues; verify time sync across CVMs and hypervisors.

Pulse (formerly CallHome) sends telemetry to Nutanix support; disabling it limits proactive support and automated health insights.

LCM (Life Cycle Manager) handles AOS, hypervisor, firmware, and NCC updates with pre-checks and staged upgrades.

Cluster capacity is reported as usable vs. raw; RF2 consumes roughly 2x the data footprint plus overhead, so plan capacity with RF in mind.

A cluster can be registered to only one Prism Central at a time; PC must reach each PE VIP over the network.

The default bridge (br0) on each host carries CVM and hypervisor management traffic; guest VM traffic uses a separate bridge or VLAN-backed network.

AHV networks are VLAN-backed virtual networks; a VM NIC attaches to a network, and the VLAN ID defines the broadcast domain.

Bond modes: active-backup (default, no switch config), balance-slb (no LACP, source MAC hashing), and balance-tcp (requires LACP on the switch).

balance-tcp requires a correctly configured LACP port-channel upstream; misconfiguration causes connectivity loss and is a common deployment error.

The cluster VIP and CVM IPs belong on a management VLAN; isolate guest VM networks from management traffic using separate VLANs.

AHV uses Open vSwitch (OVS); Flow microsegmentation policies are defined in Prism Central using categories.

Diagnose with Prism network visualization, CVM ping/arp tests, and NCC network checks; failed CVM-to-CVM ping indicates a bridge or VLAN issue.

On ESXi clusters, the CVM uses a vSwitch port group; enable promiscuous mode and forged transmits on the dedicated port group.

A storage pool groups physical disks (HDD/SSD/NVMe); a container is a logical namespace (datastore) presented to the hypervisor, and multiple containers can share a pool.

Replication factor (RF) 2 tolerates one node/disk failure; RF3 tolerates two. RF is set per container or as a cluster default and consumes capacity accordingly.

Erasure coding (EC-X) reduces capacity overhead for cold data; it is configured per container and requires a minimum number of nodes (typically 4+).

The distributed storage fabric (DSF) is log-structured and extent-based; writes land in the local CVM staging area, then replicate to peers before acknowledgment.

Deduplication (capacity and/or inline) and compression are container-level settings; compression is inline by default on supported AOS versions.

Snapshots are metadata-based and space-efficient; they are not full copies, but changed blocks consume space over time.

Storage QoS (IOPS limits) can be applied per VM or per container to prevent noisy-neighbor effects.

Check storage health via the Prism storage dashboard, NCC disk checks, and ncli; a disk in a bad state triggers a rebuild that consumes cluster resources.

In AHV, VMs are created from images (ISOs, disk images) stored in a container; images can be uploaded via Prism, ncli, or the API.

Categories (key-value pairs) drive policy application in Prism Central for protection, recovery, and security; use categories to group VMs.

AHV supports live migration, HA (restart on host failure), and affinity/anti-affinity rules enforced by the scheduler.

VM affinity rules keep VMs on the same host; anti-affinity rules spread them across hosts. Rules can be 'should' or 'must' strength.

Prism Central Projects provide self-service VM provisioning with quotas, role-based access, and environment isolation.

Nutanix Guest Tools (NGT) enables file-level restore, application-consistent snapshots, and self-service restore for supported guest OSes.

VM power operations, console access, and snapshot/clone actions are available in Prism Element; bulk operations and policies are in Prism Central.

A VM clone can be full or linked (dependent on the parent snapshot); linked clones save space but depend on the parent image/snapshot chain.

Protection domains group VMs (by category) for replication and snapshot policies; a protection policy defines RPO, retention, and replication target.

Async replication sends snapshots to a remote cluster at the RPO interval; sync replication (Metro Availability) requires low-latency links and provides near-zero RPO.

Nutanix DR uses protection policies in Prism Central; a recovery plan defines boot order and network mapping for failover.

Snapshots are local and fast; replication is remote. A protection policy can take local snapshots only, or local plus remote replication.

NearSync replication provides RPO in seconds (typically 1-5 minutes) and requires a compatible AOS version and network latency budget.

For file-level restore, NGT must be installed in the guest; self-service restore lets users recover their own files from a snapshot.

Test failover (planned) is non-disruptive; unplanned failover promotes the remote copy. Validate the recovery plan before relying on it.

Retention policies define how many snapshots/replicas to keep (e.g., hourly/daily/weekly); retention applies per protection policy and consumes capacity on both sites.

NCC is the first diagnostic step: run it from the CVM CLI or via Prism; failed checks often point to the root cause (time sync, disk, network).

CVM memory pressure is a common cause of cluster instability; monitor CVM memory and resize if consistently high.

A disk failure triggers a rebuild; monitor rebuild progress in Prism and avoid maintenance operations during a rebuild.

NTP drift causes authentication, replication, and cluster services failures; verify NTP on CVMs and hypervisors with ntpq -p or chronyc sources.

Log collection: generate a log bundle via ncli or Prism; logs are also on the CVM under /home/nutanix/data/logs.

LCM pre-checks must pass before upgrades; common blockers are insufficient disk space, NCC failures, and version incompatibility.

A node in maintenance mode evacuates VMs and data; use it before hardware service to avoid data loss and keep RF satisfied.

Cluster health is not just green lights - check the health dashboard, NCC, and capacity trends; a cluster can be functional but degraded (e.g., RF not satisfied).

Prism Central is the multicloud control plane; it manages multiple Prism Element clusters (on-prem and in public cloud) from one UI.

Nutanix Clusters on AWS/Azure extends AOS networking and management to public cloud; it is not the same as running AHV on-prem.

Calm (Nutanix Cloud Manager) provides orchestration and self-service for multi-cloud workloads; blueprints define multi-tier applications.

Nutanix Cloud Manager (NCM) includes cost governance, security, and self-service; it integrates with Prism Central and public cloud accounts.

Flow (microsegmentation) policies are defined in Prism Central and applied across clusters; they use categories, not IP addresses, for policy targets.

Multi-cluster management requires network connectivity from Prism Central to each PE VIP and consistent time sync; a cluster can be registered to only one PC.

Categories are the key mechanism for applying policies (protection, security, cost) consistently across clusters and clouds.

Cloud-native workloads can be managed alongside AHV VMs; Nutanix provides CSI/CNI integrations for Kubernetes (NKP) in multicloud scenarios.

Foundation discovers nodes via IPMI/Redfish out-of-band; it wipes disks, installs the hypervisor and AOS, and creates the cluster from a jump host.

Minimum 3 nodes for RF2, 4 nodes for RF3; single-node clusters exist but provide no redundancy.

Cluster creation requires cluster name, VIP, CVM IPs, hypervisor IPs, subnet mask, gateway, DNS, and NTP; the VIP must be a free IP in the CVM subnet.

Default CVM memory is 16 GB (32 GB for large nodes); the CVM reserves host memory, so size hosts for guest VMs plus CVM overhead.

AOS and hypervisor version compatibility is enforced by the compatibility matrix; mismatches block upgrades and cluster expansion.

After Foundation, run NCC health checks and confirm a healthy state before adding workloads; a new cluster starts with a default storage pool and container.

Prism Element manages one cluster; Prism Central is a separate VM (or scale-out set) that manages many clusters via registered Prism Element instances.

The cluster VIP, CVM IPs, and hypervisor management IPs must be unique and outside any DHCP pool used by guest VMs.

Prism Element shows cluster CPU, memory, storage, and IOPS for one cluster; Prism Central adds multi-cluster dashboards, categories, and policies.

NCC (Nutanix Cluster Check) runs a library of health checks; schedule it and triage failures - a failed check does not automatically mean an outage.

Alerts have severity levels (Critical, Warning, Info) and can route to email, SNMP, syslog, or webhooks; they can be acknowledged or resolved in Prism.

NTP drift is a common cause of cluster-wide issues; verify time sync across CVMs and hypervisors.

Pulse (formerly CallHome) sends telemetry to Nutanix support; disabling it limits proactive support and automated health insights.

LCM (Life Cycle Manager) handles AOS, hypervisor, firmware, and NCC updates with pre-checks and staged upgrades.

Cluster capacity is reported as usable vs. raw; RF2 consumes roughly 2x the data footprint plus overhead, so plan capacity with RF in mind.

A cluster can be registered to only one Prism Central at a time; PC must reach each PE VIP over the network.

The default bridge (br0) on each host carries CVM and hypervisor management traffic; guest VM traffic uses a separate bridge or VLAN-backed network.

AHV networks are VLAN-backed virtual networks; a VM NIC attaches to a network, and the VLAN ID defines the broadcast domain.

Bond modes: active-backup (default, no switch config), balance-slb (no LACP, source MAC hashing), and balance-tcp (requires LACP on the switch).

balance-tcp requires a correctly configured LACP port-channel upstream; misconfiguration causes connectivity loss and is a common deployment error.

The cluster VIP and CVM IPs belong on a management VLAN; isolate guest VM networks from management traffic using separate VLANs.

AHV uses Open vSwitch (OVS); Flow microsegmentation policies are defined in Prism Central using categories.

Diagnose with Prism network visualization, CVM ping/arp tests, and NCC network checks; failed CVM-to-CVM ping indicates a bridge or VLAN issue.

On ESXi clusters, the CVM uses a vSwitch port group; enable promiscuous mode and forged transmits on the dedicated port group.

A storage pool groups physical disks (HDD/SSD/NVMe); a container is a logical namespace (datastore) presented to the hypervisor, and multiple containers can share a pool.

Replication factor (RF) 2 tolerates one node/disk failure; RF3 tolerates two. RF is set per container or as a cluster default and consumes capacity accordingly.

Erasure coding (EC-X) reduces capacity overhead for cold data; it is configured per container and requires a minimum number of nodes (typically 4+).

The distributed storage fabric (DSF) is log-structured and extent-based; writes land in the local CVM staging area, then replicate to peers before acknowledgment.

Deduplication (capacity and/or inline) and compression are container-level settings; compression is inline by default on supported AOS versions.

Snapshots are metadata-based and space-efficient; they are not full copies, but changed blocks consume space over time.

Storage QoS (IOPS limits) can be applied per VM or per container to prevent noisy-neighbor effects.

Check storage health via the Prism storage dashboard, NCC disk checks, and ncli; a disk in a bad state triggers a rebuild that consumes cluster resources.

In AHV, VMs are created from images (ISOs, disk images) stored in a container; images can be uploaded via Prism, ncli, or the API.

Categories (key-value pairs) drive policy application in Prism Central for protection, recovery, and security; use categories to group VMs.

AHV supports live migration, HA (restart on host failure), and affinity/anti-affinity rules enforced by the scheduler.

VM affinity rules keep VMs on the same host; anti-affinity rules spread them across hosts. Rules can be 'should' or 'must' strength.

Prism Central Projects provide self-service VM provisioning with quotas, role-based access, and environment isolation.

Nutanix Guest Tools (NGT) enables file-level restore, application-consistent snapshots, and self-service restore for supported guest OSes.

VM power operations, console access, and snapshot/clone actions are available in Prism Element; bulk operations and policies are in Prism Central.

A VM clone can be full or linked (dependent on the parent snapshot); linked clones save space but depend on the parent image/snapshot chain.

Protection domains group VMs (by category) for replication and snapshot policies; a protection policy defines RPO, retention, and replication target.

Async replication sends snapshots to a remote cluster at the RPO interval; sync replication (Metro Availability) requires low-latency links and provides near-zero RPO.

Nutanix DR uses protection policies in Prism Central; a recovery plan defines boot order and network mapping for failover.

Snapshots are local and fast; replication is remote. A protection policy can take local snapshots only, or local plus remote replication.

NearSync replication provides RPO in seconds (typically 1-5 minutes) and requires a compatible AOS version and network latency budget.

For file-level restore, NGT must be installed in the guest; self-service restore lets users recover their own files from a snapshot.

Test failover (planned) is non-disruptive; unplanned failover promotes the remote copy. Validate the recovery plan before relying on it.

Retention policies define how many snapshots/replicas to keep (e.g., hourly/daily/weekly); retention applies per protection policy and consumes capacity on both sites.

NCC is the first diagnostic step: run it from the CVM CLI or via Prism; failed checks often point to the root cause (time sync, disk, network).

CVM memory pressure is a common cause of cluster instability; monitor CVM memory and resize if consistently high.

A disk failure triggers a rebuild; monitor rebuild progress in Prism and avoid maintenance operations during a rebuild.

NTP drift causes authentication, replication, and cluster services failures; verify NTP on CVMs and hypervisors with ntpq -p or chronyc sources.

Log collection: generate a log bundle via ncli or Prism; logs are also on the CVM under /home/nutanix/data/logs.

LCM pre-checks must pass before upgrades; common blockers are insufficient disk space, NCC failures, and version incompatibility.

A node in maintenance mode evacuates VMs and data; use it before hardware service to avoid data loss and keep RF satisfied.

Cluster health is not just green lights - check the health dashboard, NCC, and capacity trends; a cluster can be functional but degraded (e.g., RF not satisfied).

Prism Central is the multicloud control plane; it manages multiple Prism Element clusters (on-prem and in public cloud) from one UI.

Nutanix Clusters on AWS/Azure extends AOS networking and management to public cloud; it is not the same as running AHV on-prem.

Calm (Nutanix Cloud Manager) provides orchestration and self-service for multi-cloud workloads; blueprints define multi-tier applications.

Nutanix Cloud Manager (NCM) includes cost governance, security, and self-service; it integrates with Prism Central and public cloud accounts.

Flow (microsegmentation) policies are defined in Prism Central and applied across clusters; they use categories, not IP addresses, for policy targets.

Multi-cluster management requires network connectivity from Prism Central to each PE VIP and consistent time sync; a cluster can be registered to only one PC.

Categories are the key mechanism for applying policies (protection, security, cost) consistently across clusters and clouds.

Cloud-native workloads can be managed alongside AHV VMs; Nutanix provides CSI/CNI integrations for Kubernetes (NKP) in multicloud scenarios.

Foundation discovers nodes via IPMI/Redfish out-of-band; it wipes disks, installs the hypervisor and AOS, and creates the cluster from a jump host.

Minimum 3 nodes for RF2, 4 nodes for RF3; single-node clusters exist but provide no redundancy.

Cluster creation requires cluster name, VIP, CVM IPs, hypervisor IPs, subnet mask, gateway, DNS, and NTP; the VIP must be a free IP in the CVM subnet.

Default CVM memory is 16 GB (32 GB for large nodes); the CVM reserves host memory, so size hosts for guest VMs plus CVM overhead.

AOS and hypervisor version compatibility is enforced by the compatibility matrix; mismatches block upgrades and cluster expansion.

After Foundation, run NCC health checks and confirm a healthy state before adding workloads; a new cluster starts with a default storage pool and container.

Prism Element manages one cluster; Prism Central is a separate VM (or scale-out set) that manages many clusters via registered Prism Element instances.

The cluster VIP, CVM IPs, and hypervisor management IPs must be unique and outside any DHCP pool used by guest VMs.

Prism Element shows cluster CPU, memory, storage, and IOPS for one cluster; Prism Central adds multi-cluster dashboards, categories, and policies.

NCC (Nutanix Cluster Check) runs a library of health checks; schedule it and triage failures - a failed check does not automatically mean an outage.

Alerts have severity levels (Critical, Warning, Info) and can route to email, SNMP, syslog, or webhooks; they can be acknowledged or resolved in Prism.

NTP drift is a common cause of cluster-wide issues; verify time sync across CVMs and hypervisors.

Pulse (formerly CallHome) sends telemetry to Nutanix support; disabling it limits proactive support and automated health insights.

LCM (Life Cycle Manager) handles AOS, hypervisor, firmware, and NCC updates with pre-checks and staged upgrades.

Cluster capacity is reported as usable vs. raw; RF2 consumes roughly 2x the data footprint plus overhead, so plan capacity with RF in mind.

A cluster can be registered to only one Prism Central at a time; PC must reach each PE VIP over the network.

The default bridge (br0) on each host carries CVM and hypervisor management traffic; guest VM traffic uses a separate bridge or VLAN-backed network.

AHV networks are VLAN-backed virtual networks; a VM NIC attaches to a network, and the VLAN ID defines the broadcast domain.

Bond modes: active-backup (default, no switch config), balance-slb (no LACP, source MAC hashing), and balance-tcp (requires LACP on the switch).

balance-tcp requires a correctly configured LACP port-channel upstream; misconfiguration causes connectivity loss and is a common deployment error.

The cluster VIP and CVM IPs belong on a management VLAN; isolate guest VM networks from management traffic using separate VLANs.

AHV uses Open vSwitch (OVS); Flow microsegmentation policies are defined in Prism Central using categories.

Diagnose with Prism network visualization, CVM ping/arp tests, and NCC network checks; failed CVM-to-CVM ping indicates a bridge or VLAN issue.

On ESXi clusters, the CVM uses a vSwitch port group; enable promiscuous mode and forged transmits on the dedicated port group.

A storage pool groups physical disks (HDD/SSD/NVMe); a container is a logical namespace (datastore) presented to the hypervisor, and multiple containers can share a pool.

Replication factor (RF) 2 tolerates one node/disk failure; RF3 tolerates two. RF is set per container or as a cluster default and consumes capacity accordingly.

Erasure coding (EC-X) reduces capacity overhead for cold data; it is configured per container and requires a minimum number of nodes (typically 4+).

The distributed storage fabric (DSF) is log-structured and extent-based; writes land in the local CVM staging area, then replicate to peers before acknowledgment.

Deduplication (capacity and/or inline) and compression are container-level settings; compression is inline by default on supported AOS versions.

Snapshots are metadata-based and space-efficient; they are not full copies, but changed blocks consume space over time.

Storage QoS (IOPS limits) can be applied per VM or per container to prevent noisy-neighbor effects.

Check storage health via the Prism storage dashboard, NCC disk checks, and ncli; a disk in a bad state triggers a rebuild that consumes cluster resources.

In AHV, VMs are created from images (ISOs, disk images) stored in a container; images can be uploaded via Prism, ncli, or the API.

Categories (key-value pairs) drive policy application in Prism Central for protection, recovery, and security; use categories to group VMs.

AHV supports live migration, HA (restart on host failure), and affinity/anti-affinity rules enforced by the scheduler.

VM affinity rules keep VMs on the same host; anti-affinity rules spread them across hosts. Rules can be 'should' or 'must' strength.

Prism Central Projects provide self-service VM provisioning with quotas, role-based access, and environment isolation.

Nutanix Guest Tools (NGT) enables file-level restore, application-consistent snapshots, and self-service restore for supported guest OSes.

VM power operations, console access, and snapshot/clone actions are available in Prism Element; bulk operations and policies are in Prism Central.

A VM clone can be full or linked (dependent on the parent snapshot); linked clones save space but depend on the parent image/snapshot chain.

Protection domains group VMs (by category) for replication and snapshot policies; a protection policy defines RPO, retention, and replication target.

Async replication sends snapshots to a remote cluster at the RPO interval; sync replication (Metro Availability) requires low-latency links and provides near-zero RPO.

Nutanix DR uses protection policies in Prism Central; a recovery plan defines boot order and network mapping for failover.

Snapshots are local and fast; replication is remote. A protection policy can take local snapshots only, or local plus remote replication.

NearSync replication provides RPO in seconds (typically 1-5 minutes) and requires a compatible AOS version and network latency budget.

For file-level restore, NGT must be installed in the guest; self-service restore lets users recover their own files from a snapshot.

Test failover (planned) is non-disruptive; unplanned failover promotes the remote copy. Validate the recovery plan before relying on it.

Retention policies define how many snapshots/replicas to keep (e.g., hourly/daily/weekly); retention applies per protection policy and consumes capacity on both sites.

NCC is the first diagnostic step: run it from the CVM CLI or via Prism; failed checks often point to the root cause (time sync, disk, network).

CVM memory pressure is a common cause of cluster instability; monitor CVM memory and resize if consistently high.

A disk failure triggers a rebuild; monitor rebuild progress in Prism and avoid maintenance operations during a rebuild.

NTP drift causes authentication, replication, and cluster services failures; verify NTP on CVMs and hypervisors with ntpq -p or chronyc sources.

Log collection: generate a log bundle via ncli or Prism; logs are also on the CVM under /home/nutanix/data/logs.

LCM pre-checks must pass before upgrades; common blockers are insufficient disk space, NCC failures, and version incompatibility.

A node in maintenance mode evacuates VMs and data; use it before hardware service to avoid data loss and keep RF satisfied.

Cluster health is not just green lights - check the health dashboard, NCC, and capacity trends; a cluster can be functional but degraded (e.g., RF not satisfied).

Prism Central is the multicloud control plane; it manages multiple Prism Element clusters (on-prem and in public cloud) from one UI.

Nutanix Clusters on AWS/Azure extends AOS networking and management to public cloud; it is not the same as running AHV on-prem.

Calm (Nutanix Cloud Manager) provides orchestration and self-service for multi-cloud workloads; blueprints define multi-tier applications.

Nutanix Cloud Manager (NCM) includes cost governance, security, and self-service; it integrates with Prism Central and public cloud accounts.

Flow (microsegmentation) policies are defined in Prism Central and applied across clusters; they use categories, not IP addresses, for policy targets.

Multi-cluster management requires network connectivity from Prism Central to each PE VIP and consistent time sync; a cluster can be registered to only one PC.

Categories are the key mechanism for applying policies (protection, security, cost) consistently across clusters and clouds.

Cloud-native workloads can be managed alongside AHV VMs; Nutanix provides CSI/CNI integrations for Kubernetes (NKP) in multicloud scenarios.

Foundation discovers nodes via IPMI/Redfish out-of-band; it wipes disks, installs the hypervisor and AOS, and creates the cluster from a jump host.

Minimum 3 nodes for RF2, 4 nodes for RF3; single-node clusters exist but provide no redundancy.

Cluster creation requires cluster name, VIP, CVM IPs, hypervisor IPs, subnet mask, gateway, DNS, and NTP; the VIP must be a free IP in the CVM subnet.

Default CVM memory is 16 GB (32 GB for large nodes); the CVM reserves host memory, so size hosts for guest VMs plus CVM overhead.

AOS and hypervisor version compatibility is enforced by the compatibility matrix; mismatches block upgrades and cluster expansion.

After Foundation, run NCC health checks and confirm a healthy state before adding workloads; a new cluster starts with a default storage pool and container.

Prism Element manages one cluster; Prism Central is a separate VM (or scale-out set) that manages many clusters via registered Prism Element instances.

The cluster VIP, CVM IPs, and hypervisor management IPs must be unique and outside any DHCP pool used by guest VMs.

Prism Element shows cluster CPU, memory, storage, and IOPS for one cluster; Prism Central adds multi-cluster dashboards, categories, and policies.

NCC (Nutanix Cluster Check) runs a library of health checks; schedule it and triage failures - a failed check does not automatically mean an outage.

Alerts have severity levels (Critical, Warning, Info) and can route to email, SNMP, syslog, or webhooks; they can be acknowledged or resolved in Prism.

NTP drift is a common cause of cluster-wide issues; verify time sync across CVMs and hypervisors.

Pulse (formerly CallHome) sends telemetry to Nutanix support; disabling it limits proactive support and automated health insights.

LCM (Life Cycle Manager) handles AOS, hypervisor, firmware, and NCC updates with pre-checks and staged upgrades.

Cluster capacity is reported as usable vs. raw; RF2 consumes roughly 2x the data footprint plus overhead, so plan capacity with RF in mind.

A cluster can be registered to only one Prism Central at a time; PC must reach each PE VIP over the network.

The default bridge (br0) on each host carries CVM and hypervisor management traffic; guest VM traffic uses a separate bridge or VLAN-backed network.

AHV networks are VLAN-backed virtual networks; a VM NIC attaches to a network, and the VLAN ID defines the broadcast domain.

Bond modes: active-backup (default, no switch config), balance-slb (no LACP, source MAC hashing), and balance-tcp (requires LACP on the switch).

balance-tcp requires a correctly configured LACP port-channel upstream; misconfiguration causes connectivity loss and is a common deployment error.

The cluster VIP and CVM IPs belong on a management VLAN; isolate guest VM networks from management traffic using separate VLANs.

AHV uses Open vSwitch (OVS); Flow microsegmentation policies are defined in Prism Central using categories.

Diagnose with Prism network visualization, CVM ping/arp tests, and NCC network checks; failed CVM-to-CVM ping indicates a bridge or VLAN issue.

On ESXi clusters, the CVM uses a vSwitch port group; enable promiscuous mode and forged transmits on the dedicated port group.

A storage pool groups physical disks (HDD/SSD/NVMe); a container is a logical namespace (datastore) presented to the hypervisor, and multiple containers can share a pool.

Replication factor (RF) 2 tolerates one node/disk failure; RF3 tolerates two. RF is set per container or as a cluster default and consumes capacity accordingly.

Erasure coding (EC-X) reduces capacity overhead for cold data; it is configured per container and requires a minimum number of nodes (typically 4+).

The distributed storage fabric (DSF) is log-structured and extent-based; writes land in the local CVM staging area, then replicate to peers before acknowledgment.

Deduplication (capacity and/or inline) and compression are container-level settings; compression is inline by default on supported AOS versions.

Snapshots are metadata-based and space-efficient; they are not full copies, but changed blocks consume space over time.

Storage QoS (IOPS limits) can be applied per VM or per container to prevent noisy-neighbor effects.

Check storage health via the Prism storage dashboard, NCC disk checks, and ncli; a disk in a bad state triggers a rebuild that consumes cluster resources.

In AHV, VMs are created from images (ISOs, disk images) stored in a container; images can be uploaded via Prism, ncli, or the API.

Categories (key-value pairs) drive policy application in Prism Central for protection, recovery, and security; use categories to group VMs.

AHV supports live migration, HA (restart on host failure), and affinity/anti-affinity rules enforced by the scheduler.

VM affinity rules keep VMs on the same host; anti-affinity rules spread them across hosts. Rules can be 'should' or 'must' strength.

Prism Central Projects provide self-service VM provisioning with quotas, role-based access, and environment isolation.

Nutanix Guest Tools (NGT) enables file-level restore, application-consistent snapshots, and self-service restore for supported guest OSes.

VM power operations, console access, and snapshot/clone actions are available in Prism Element; bulk operations and policies are in Prism Central.

A VM clone can be full or linked (dependent on the parent snapshot); linked clones save space but depend on the parent image/snapshot chain.

Protection domains group VMs (by category) for replication and snapshot policies; a protection policy defines RPO, retention, and replication target.

Async replication sends snapshots to a remote cluster at the RPO interval; sync replication (Metro Availability) requires low-latency links and provides near-zero RPO.

Nutanix DR uses protection policies in Prism Central; a recovery plan defines boot order and network mapping for failover.

Snapshots are local and fast; replication is remote. A protection policy can take local snapshots only, or local plus remote replication.

NearSync replication provides RPO in seconds (typically 1-5 minutes) and requires a compatible AOS version and network latency budget.

For file-level restore, NGT must be installed in the guest; self-service restore lets users recover their own files from a snapshot.

Test failover (planned) is non-disruptive; unplanned failover promotes the remote copy. Validate the recovery plan before relying on it.

Retention policies define how many snapshots/replicas to keep (e.g., hourly/daily/weekly); retention applies per protection policy and consumes capacity on both sites.

NCC is the first diagnostic step: run it from the CVM CLI or via Prism; failed checks often point to the root cause (time sync, disk, network).

CVM memory pressure is a common cause of cluster instability; monitor CVM memory and resize if consistently high.

A disk failure triggers a rebuild; monitor rebuild progress in Prism and avoid maintenance operations during a rebuild.

NTP drift causes authentication, replication, and cluster services failures; verify NTP on CVMs and hypervisors with ntpq -p or chronyc sources.

Log collection: generate a log bundle via ncli or Prism; logs are also on the CVM under /home/nutanix/data/logs.

LCM pre-checks must pass before upgrades; common blockers are insufficient disk space, NCC failures, and version incompatibility.

A node in maintenance mode evacuates VMs and data; use it before hardware service to avoid data loss and keep RF satisfied.

Cluster health is not just green lights - check the health dashboard, NCC, and capacity trends; a cluster can be functional but degraded (e.g., RF not satisfied).

Prism Central is the multicloud control plane; it manages multiple Prism Element clusters (on-prem and in public cloud) from one UI.

Nutanix Clusters on AWS/Azure extends AOS networking and management to public cloud; it is not the same as running AHV on-prem.

Calm (Nutanix Cloud Manager) provides orchestration and self-service for multi-cloud workloads; blueprints define multi-tier applications.

Nutanix Cloud Manager (NCM) includes cost governance, security, and self-service; it integrates with Prism Central and public cloud accounts.

Flow (microsegmentation) policies are defined in Prism Central and applied across clusters; they use categories, not IP addresses, for policy targets.

Multi-cluster management requires network connectivity from Prism Central to each PE VIP and consistent time sync; a cluster can be registered to only one PC.

Categories are the key mechanism for applying policies (protection, security, cost) consistently across clusters and clouds.

Cloud-native workloads can be managed alongside AHV VMs; Nutanix provides CSI/CNI integrations for Kubernetes (NKP) in multicloud scenarios.

Foundation discovers nodes via IPMI/Redfish out-of-band; it wipes disks, installs the hypervisor and AOS, and creates the cluster from a jump host.

Minimum 3 nodes for RF2, 4 nodes for RF3; single-node clusters exist but provide no redundancy.

Cluster creation requires cluster name, VIP, CVM IPs, hypervisor IPs, subnet mask, gateway, DNS, and NTP; the VIP must be a free IP in the CVM subnet.

Default CVM memory is 16 GB (32 GB for large nodes); the CVM reserves host memory, so size hosts for guest VMs plus CVM overhead.

AOS and hypervisor version compatibility is enforced by the compatibility matrix; mismatches block upgrades and cluster expansion.

After Foundation, run NCC health checks and confirm a healthy state before adding workloads; a new cluster starts with a default storage pool and container.

Prism Element manages one cluster; Prism Central is a separate VM (or scale-out set) that manages many clusters via registered Prism Element instances.

The cluster VIP, CVM IPs, and hypervisor management IPs must be unique and outside any DHCP pool used by guest VMs.

Prism Element shows cluster CPU, memory, storage, and IOPS for one cluster; Prism Central adds multi-cluster dashboards, categories, and policies.

NCC (Nutanix Cluster Check) runs a library of health checks; schedule it and triage failures - a failed check does not automatically mean an outage.

Alerts have severity levels (Critical, Warning, Info) and can route to email, SNMP, syslog, or webhooks; they can be acknowledged or resolved in Prism.

NTP drift is a common cause of cluster-wide issues; verify time sync across CVMs and hypervisors.

Pulse (formerly CallHome) sends telemetry to Nutanix support; disabling it limits proactive support and automated health insights.

LCM (Life Cycle Manager) handles AOS, hypervisor, firmware, and NCC updates with pre-checks and staged upgrades.

Cluster capacity is reported as usable vs. raw; RF2 consumes roughly 2x the data footprint plus overhead, so plan capacity with RF in mind.

A cluster can be registered to only one Prism Central at a time; PC must reach each PE VIP over the network.

The default bridge (br0) on each host carries CVM and hypervisor management traffic; guest VM traffic uses a separate bridge or VLAN-backed network.

AHV networks are VLAN-backed virtual networks; a VM NIC attaches to a network, and the VLAN ID defines the broadcast domain.

Bond modes: active-backup (default, no switch config), balance-slb (no LACP, source MAC hashing), and balance-tcp (requires LACP on the switch).

balance-tcp requires a correctly configured LACP port-channel upstream; misconfiguration causes connectivity loss and is a common deployment error.

The cluster VIP and CVM IPs belong on a management VLAN; isolate guest VM networks from management traffic using separate VLANs.

AHV uses Open vSwitch (OVS); Flow microsegmentation policies are defined in Prism Central using categories.

Diagnose with Prism network visualization, CVM ping/arp tests, and NCC network checks; failed CVM-to-CVM ping indicates a bridge or VLAN issue.

On ESXi clusters, the CVM uses a vSwitch port group; enable promiscuous mode and forged transmits on the dedicated port group.

A storage pool groups physical disks (HDD/SSD/NVMe); a container is a logical namespace (datastore) presented to the hypervisor, and multiple containers can share a pool.

Replication factor (RF) 2 tolerates one node/disk failure; RF3 tolerates two. RF is set per container or as a cluster default and consumes capacity accordingly.

Erasure coding (EC-X) reduces capacity overhead for cold data; it is configured per container and requires a minimum number of nodes (typically 4+).

The distributed storage fabric (DSF) is log-structured and extent-based; writes land in the local CVM staging area, then replicate to peers before acknowledgment.

Deduplication (capacity and/or inline) and compression are container-level settings; compression is inline by default on supported AOS versions.

Snapshots are metadata-based and space-efficient; they are not full copies, but changed blocks consume space over time.

Storage QoS (IOPS limits) can be applied per VM or per container to prevent noisy-neighbor effects.

Check storage health via the Prism storage dashboard, NCC disk checks, and ncli; a disk in a bad state triggers a rebuild that consumes cluster resources.

In AHV, VMs are created from images (ISOs, disk images) stored in a container; images can be uploaded via Prism, ncli, or the API.

Categories (key-value pairs) drive policy application in Prism Central for protection, recovery, and security; use categories to group VMs.

AHV supports live migration, HA (restart on host failure), and affinity/anti-affinity rules enforced by the scheduler.

VM affinity rules keep VMs on the same host; anti-affinity rules spread them across hosts. Rules can be 'should' or 'must' strength.

Prism Central Projects provide self-service VM provisioning with quotas, role-based access, and environment isolation.

Nutanix Guest Tools (NGT) enables file-level restore, application-consistent snapshots, and self-service restore for supported guest OSes.

VM power operations, console access, and snapshot/clone actions are available in Prism Element; bulk operations and policies are in Prism Central.

A VM clone can be full or linked (dependent on the parent snapshot); linked clones save space but depend on the parent image/snapshot chain.

Protection domains group VMs (by category) for replication and snapshot policies; a protection policy defines RPO, retention, and replication target.

Async replication sends snapshots to a remote cluster at the RPO interval; sync replication (Metro Availability) requires low-latency links and provides near-zero RPO.

Nutanix DR uses protection policies in Prism Central; a recovery plan defines boot order and network mapping for failover.

Snapshots are local and fast; replication is remote. A protection policy can take local snapshots only, or local plus remote replication.

NearSync replication provides RPO in seconds (typically 1-5 minutes) and requires a compatible AOS version and network latency budget.

For file-level restore, NGT must be installed in the guest; self-service restore lets users recover their own files from a snapshot.

Test failover (planned) is non-disruptive; unplanned failover promotes the remote copy. Validate the recovery plan before relying on it.

Retention policies define how many snapshots/replicas to keep (e.g., hourly/daily/weekly); retention applies per protection policy and consumes capacity on both sites.

NCC is the first diagnostic step: run it from the CVM CLI or via Prism; failed checks often point to the root cause (time sync, disk, network).

CVM memory pressure is a common cause of cluster instability; monitor CVM memory and resize if consistently high.

A disk failure triggers a rebuild; monitor rebuild progress in Prism and avoid maintenance operations during a rebuild.

NTP drift causes authentication, replication, and cluster services failures; verify NTP on CVMs and hypervisors with ntpq -p or chronyc sources.

Log collection: generate a log bundle via ncli or Prism; logs are also on the CVM under /home/nutanix/data/logs.

LCM pre-checks must pass before upgrades; common blockers are insufficient disk space, NCC failures, and version incompatibility.

A node in maintenance mode evacuates VMs and data; use it before hardware service to avoid data loss and keep RF satisfied.

Cluster health is not just green lights - check the health dashboard, NCC, and capacity trends; a cluster can be functional but degraded (e.g., RF not satisfied).

Prism Central is the multicloud control plane; it manages multiple Prism Element clusters (on-prem and in public cloud) from one UI.

Nutanix Clusters on AWS/Azure extends AOS networking and management to public cloud; it is not the same as running AHV on-prem.

Calm (Nutanix Cloud Manager) provides orchestration and self-service for multi-cloud workloads; blueprints define multi-tier applications.

Nutanix Cloud Manager (NCM) includes cost governance, security, and self-service; it integrates with Prism Central and public cloud accounts.

Flow (microsegmentation) policies are defined in Prism Central and applied across clusters; they use categories, not IP addresses, for policy targets.

Multi-cluster management requires network connectivity from Prism Central to each PE VIP and consistent time sync; a cluster can be registered to only one PC.

Categories are the key mechanism for applying policies (protection, security, cost) consistently across clusters and clouds.

Cloud-native workloads can be managed alongside AHV VMs; Nutanix provides CSI/CNI integrations for Kubernetes (NKP) in multicloud scenarios.

Foundation discovers nodes via IPMI/Redfish out-of-band; it wipes disks, installs the hypervisor and AOS, and creates the cluster from a jump host.

Minimum 3 nodes for RF2, 4 nodes for RF3; single-node clusters exist but provide no redundancy.

Cluster creation requires cluster name, VIP, CVM IPs, hypervisor IPs, subnet mask, gateway, DNS, and NTP; the VIP must be a free IP in the CVM subnet.

Default CVM memory is 16 GB (32 GB for large nodes); the CVM reserves host memory, so size hosts for guest VMs plus CVM overhead.

AOS and hypervisor version compatibility is enforced by the compatibility matrix; mismatches block upgrades and cluster expansion.

After Foundation, run NCC health checks and confirm a healthy state before adding workloads; a new cluster starts with a default storage pool and container.

Prism Element manages one cluster; Prism Central is a separate VM (or scale-out set) that manages many clusters via registered Prism Element instances.

The cluster VIP, CVM IPs, and hypervisor management IPs must be unique and outside any DHCP pool used by guest VMs.

Prism Element shows cluster CPU, memory, storage, and IOPS for one cluster; Prism Central adds multi-cluster dashboards, categories, and policies.

NCC (Nutanix Cluster Check) runs a library of health checks; schedule it and triage failures - a failed check does not automatically mean an outage.

Alerts have severity levels (Critical, Warning, Info) and can route to email, SNMP, syslog, or webhooks; they can be acknowledged or resolved in Prism.

NTP drift is a common cause of cluster-wide issues; verify time sync across CVMs and hypervisors.

Pulse (formerly CallHome) sends telemetry to Nutanix support; disabling it limits proactive support and automated health insights.

LCM (Life Cycle Manager) handles AOS, hypervisor, firmware, and NCC updates with pre-checks and staged upgrades.

Cluster capacity is reported as usable vs. raw; RF2 consumes roughly 2x the data footprint plus overhead, so plan capacity with RF in mind.

A cluster can be registered to only one Prism Central at a time; PC must reach each PE VIP over the network.

The default bridge (br0) on each host carries CVM and hypervisor management traffic; guest VM traffic uses a separate bridge or VLAN-backed network.

AHV networks are VLAN-backed virtual networks; a VM NIC attaches to a network, and the VLAN ID defines the broadcast domain.

Bond modes: active-backup (default, no switch config), balance-slb (no LACP, source MAC hashing), and balance-tcp (requires LACP on the switch).

balance-tcp requires a correctly configured LACP port-channel upstream; misconfiguration causes connectivity loss and is a common deployment error.

The cluster VIP and CVM IPs belong on a management VLAN; isolate guest VM networks from management traffic using separate VLANs.

AHV uses Open vSwitch (OVS); Flow microsegmentation policies are defined in Prism Central using categories.

Diagnose with Prism network visualization, CVM ping/arp tests, and NCC network checks; failed CVM-to-CVM ping indicates a bridge or VLAN issue.

On ESXi clusters, the CVM uses a vSwitch port group; enable promiscuous mode and forged transmits on the dedicated port group.

A storage pool groups physical disks (HDD/SSD/NVMe); a container is a logical namespace (datastore) presented to the hypervisor, and multiple containers can share a pool.

Replication factor (RF) 2 tolerates one node/disk failure; RF3 tolerates two. RF is set per container or as a cluster default and consumes capacity accordingly.

Erasure coding (EC-X) reduces capacity overhead for cold data; it is configured per container and requires a minimum number of nodes (typically 4+).

The distributed storage fabric (DSF) is log-structured and extent-based; writes land in the local CVM staging area, then replicate to peers before acknowledgment.

Deduplication (capacity and/or inline) and compression are container-level settings; compression is inline by default on supported AOS versions.

Snapshots are metadata-based and space-efficient; they are not full copies, but changed blocks consume space over time.

Storage QoS (IOPS limits) can be applied per VM or per container to prevent noisy-neighbor effects.

Check storage health via the Prism storage dashboard, NCC disk checks, and ncli; a disk in a bad state triggers a rebuild that consumes cluster resources.

In AHV, VMs are created from images (ISOs, disk images) stored in a container; images can be uploaded via Prism, ncli, or the API.

Categories (key-value pairs) drive policy application in Prism Central for protection, recovery, and security; use categories to group VMs.

AHV supports live migration, HA (restart on host failure), and affinity/anti-affinity rules enforced by the scheduler.

VM affinity rules keep VMs on the same host; anti-affinity rules spread them across hosts. Rules can be 'should' or 'must' strength.

Prism Central Projects provide self-service VM provisioning with quotas, role-based access, and environment isolation.

Nutanix Guest Tools (NGT) enables file-level restore, application-consistent snapshots, and self-service restore for supported guest OSes.

VM power operations, console access, and snapshot/clone actions are available in Prism Element; bulk operations and policies are in Prism Central.

A VM clone can be full or linked (dependent on the parent snapshot); linked clones save space but depend on the parent image/snapshot chain.

Protection domains group VMs (by category) for replication and snapshot policies; a protection policy defines RPO, retention, and replication target.

Async replication sends snapshots to a remote cluster at the RPO interval; sync replication (Metro Availability) requires low-latency links and provides near-zero RPO.

Nutanix DR uses protection policies in Prism Central; a recovery plan defines boot order and network mapping for failover.

Snapshots are local and fast; replication is remote. A protection policy can take local snapshots only, or local plus remote replication.

NearSync replication provides RPO in seconds (typically 1-5 minutes) and requires a compatible AOS version and network latency budget.

For file-level restore, NGT must be installed in the guest; self-service restore lets users recover their own files from a snapshot.

Test failover (planned) is non-disruptive; unplanned failover promotes the remote copy. Validate the recovery plan before relying on it.

Retention policies define how many snapshots/replicas to keep (e.g., hourly/daily/weekly); retention applies per protection policy and consumes capacity on both sites.

NCC is the first diagnostic step: run it from the CVM CLI or via Prism; failed checks often point to the root cause (time sync, disk, network).

CVM memory pressure is a common cause of cluster instability; monitor CVM memory and resize if consistently high.

A disk failure triggers a rebuild; monitor rebuild progress in Prism and avoid maintenance operations during a rebuild.

NTP drift causes authentication, replication, and cluster services failures; verify NTP on CVMs and hypervisors with ntpq -p or chronyc sources.

Log collection: generate a log bundle via ncli or Prism; logs are also on the CVM under /home/nutanix/data/logs.

LCM pre-checks must pass before upgrades; common blockers are insufficient disk space, NCC failures, and version incompatibility.

A node in maintenance mode evacuates VMs and data; use it before hardware service to avoid data loss and keep RF satisfied.

Cluster health is not just green lights - check the health dashboard, NCC, and capacity trends; a cluster can be functional but degraded (e.g., RF not satisfied).

Prism Central is the multicloud control plane; it manages multiple Prism Element clusters (on-prem and in public cloud) from one UI.

Nutanix Clusters on AWS/Azure extends AOS networking and management to public cloud; it is not the same as running AHV on-prem.

Calm (Nutanix Cloud Manager) provides orchestration and self-service for multi-cloud workloads; blueprints define multi-tier applications.

Nutanix Cloud Manager (NCM) includes cost governance, security, and self-service; it integrates with Prism Central and public cloud accounts.

Flow (microsegmentation) policies are defined in Prism Central and applied across clusters; they use categories, not IP addresses, for policy targets.

Multi-cluster management requires network connectivity from Prism Central to each PE VIP and consistent time sync; a cluster can be registered to only one PC.

Categories are the key mechanism for applying policies (protection, security, cost) consistently across clusters and clouds.

Cloud-native workloads can be managed alongside AHV VMs; Nutanix provides CSI/CNI integrations for Kubernetes (NKP) in multicloud scenarios.

Foundation discovers nodes via IPMI/Redfish out-of-band; it wipes disks, installs the hypervisor and AOS, and creates the cluster from a jump host.

Minimum 3 nodes for RF2, 4 nodes for RF3; single-node clusters exist but provide no redundancy.

Cluster creation requires cluster name, VIP, CVM IPs, hypervisor IPs, subnet mask, gateway, DNS, and NTP; the VIP must be a free IP in the CVM subnet.

Default CVM memory is 16 GB (32 GB for large nodes); the CVM reserves host memory, so size hosts for guest VMs plus CVM overhead.

AOS and hypervisor version compatibility is enforced by the compatibility matrix; mismatches block upgrades and cluster expansion.

After Foundation, run NCC health checks and confirm a healthy state before adding workloads; a new cluster starts with a default storage pool and container.

Prism Element manages one cluster; Prism Central is a separate VM (or scale-out set) that manages many clusters via registered Prism Element instances.

The cluster VIP, CVM IPs, and hypervisor management IPs must be unique and outside any DHCP pool used by guest VMs.

Prism Element shows cluster CPU, memory, storage, and IOPS for one cluster; Prism Central adds multi-cluster dashboards, categories, and policies.

NCC (Nutanix Cluster Check) runs a library of health checks; schedule it and triage failures - a failed check does not automatically mean an outage.

Alerts have severity levels (Critical, Warning, Info) and can route to email, SNMP, syslog, or webhooks; they can be acknowledged or resolved in Prism.

NTP drift is a common cause of cluster-wide issues; verify time sync across CVMs and hypervisors.

Pulse (formerly CallHome) sends telemetry to Nutanix support; disabling it limits proactive support and automated health insights.

LCM (Life Cycle Manager) handles AOS, hypervisor, firmware, and NCC updates with pre-checks and staged upgrades.

Cluster capacity is reported as usable vs. raw; RF2 consumes roughly 2x the data footprint plus overhead, so plan capacity with RF in mind.

A cluster can be registered to only one Prism Central at a time; PC must reach each PE VIP over the network.

The default bridge (br0) on each host carries CVM and hypervisor management traffic; guest VM traffic uses a separate bridge or VLAN-backed network.

AHV networks are VLAN-backed virtual networks; a VM NIC attaches to a network, and the VLAN ID defines the broadcast domain.

Bond modes: active-backup (default, no switch config), balance-slb (no LACP, source MAC hashing), and balance-tcp (requires LACP on the switch).

balance-tcp requires a correctly configured LACP port-channel upstream; misconfiguration causes connectivity loss and is a common deployment error.

The cluster VIP and CVM IPs belong on a management VLAN; isolate guest VM networks from management traffic using separate VLANs.

AHV uses Open vSwitch (OVS); Flow microsegmentation policies are defined in Prism Central using categories.

Diagnose with Prism network visualization, CVM ping/arp tests, and NCC network checks; failed CVM-to-CVM ping indicates a bridge or VLAN issue.

On ESXi clusters, the CVM uses a vSwitch port group; enable promiscuous mode and forged transmits on the dedicated port group.

A storage pool groups physical disks (HDD/SSD/NVMe); a container is a logical namespace (datastore) presented to the hypervisor, and multiple containers can share a pool.

Replication factor (RF) 2 tolerates one node/disk failure; RF3 tolerates two. RF is set per container or as a cluster default and consumes capacity accordingly.

Erasure coding (EC-X) reduces capacity overhead for cold data; it is configured per container and requires a minimum number of nodes (typically 4+).

The distributed storage fabric (DSF) is log-structured and extent-based; writes land in the local CVM staging area, then replicate to peers before acknowledgment.

Deduplication (capacity and/or inline) and compression are container-level settings; compression is inline by default on supported AOS versions.

Snapshots are metadata-based and space-efficient; they are not full copies, but changed blocks consume space over time.

Storage QoS (IOPS limits) can be applied per VM or per container to prevent noisy-neighbor effects.

Check storage health via the Prism storage dashboard, NCC disk checks, and ncli; a disk in a bad state triggers a rebuild that consumes cluster resources.

In AHV, VMs are created from images (ISOs, disk images) stored in a container; images can be uploaded via Prism, ncli, or the API.

Categories (key-value pairs) drive policy application in Prism Central for protection, recovery, and security; use categories to group VMs.

AHV supports live migration, HA (restart on host failure), and affinity/anti-affinity rules enforced by the scheduler.

VM affinity rules keep VMs on the same host; anti-affinity rules spread them across hosts. Rules can be 'should' or 'must' strength.

Prism Central Projects provide self-service VM provisioning with quotas, role-based access, and environment isolation.

Nutanix Guest Tools (NGT) enables file-level restore, application-consistent snapshots, and self-service restore for supported guest OSes.

VM power operations, console access, and snapshot/clone actions are available in Prism Element; bulk operations and policies are in Prism Central.

A VM clone can be full or linked (dependent on the parent snapshot); linked clones save space but depend on the parent image/snapshot chain.

Protection domains group VMs (by category) for replication and snapshot policies; a protection policy defines RPO, retention, and replication target.

Async replication sends snapshots to a remote cluster at the RPO interval; sync replication (Metro Availability) requires low-latency links and provides near-zero RPO.

Nutanix DR uses protection policies in Prism Central; a recovery plan defines boot order and network mapping for failover.

Snapshots are local and fast; replication is remote. A protection policy can take local snapshots only, or local plus remote replication.

NearSync replication provides RPO in seconds (typically 1-5 minutes) and requires a compatible AOS version and network latency budget.

For file-level restore, NGT must be installed in the guest; self-service restore lets users recover their own files from a snapshot.

Test failover (planned) is non-disruptive; unplanned failover promotes the remote copy. Validate the recovery plan before relying on it.

Retention policies define how many snapshots/replicas to keep (e.g., hourly/daily/weekly); retention applies per protection policy and consumes capacity on both sites.

NCC is the first diagnostic step: run it from the CVM CLI or via Prism; failed checks often point to the root cause (time sync, disk, network).

CVM memory pressure is a common cause of cluster instability; monitor CVM memory and resize if consistently high.

A disk failure triggers a rebuild; monitor rebuild progress in Prism and avoid maintenance operations during a rebuild.

NTP drift causes authentication, replication, and cluster services failures; verify NTP on CVMs and hypervisors with ntpq -p or chronyc sources.

Log collection: generate a log bundle via ncli or Prism; logs are also on the CVM under /home/nutanix/data/logs.

LCM pre-checks must pass before upgrades; common blockers are insufficient disk space, NCC failures, and version incompatibility.

A node in maintenance mode evacuates VMs and data; use it before hardware service to avoid data loss and keep RF satisfied.

Cluster health is not just green lights - check the health dashboard, NCC, and capacity trends; a cluster can be functional but degraded (e.g., RF not satisfied).

Prism Central is the multicloud control plane; it manages multiple Prism Element clusters (on-prem and in public cloud) from one UI.

Nutanix Clusters on AWS/Azure extends AOS networking and management to public cloud; it is not the same as running AHV on-prem.

Calm (Nutanix Cloud Manager) provides orchestration and self-service for multi-cloud workloads; blueprints define multi-tier applications.

Nutanix Cloud Manager (NCM) includes cost governance, security, and self-service; it integrates with Prism Central and public cloud accounts.

Flow (microsegmentation) policies are defined in Prism Central and applied across clusters; they use categories, not IP addresses, for policy targets.

Multi-cluster management requires network connectivity from Prism Central to each PE VIP and consistent time sync; a cluster can be registered to only one PC.

Categories are the key mechanism for applying policies (protection, security, cost) consistently across clusters and clouds.

Cloud-native workloads can be managed alongside AHV VMs; Nutanix provides CSI/CNI integrations for Kubernetes (NKP) in multicloud scenarios.

Foundation discovers nodes via IPMI/Redfish out-of-band; it wipes disks, installs the hypervisor and AOS, and creates the cluster from a jump host.

Minimum 3 nodes for RF2, 4 nodes for RF3; single-node clusters exist but provide no redundancy.

Cluster creation requires cluster name, VIP, CVM IPs, hypervisor IPs, subnet mask, gateway, DNS, and NTP; the VIP must be a free IP in the CVM subnet.

Default CVM memory is 16 GB (32 GB for large nodes); the CVM reserves host memory, so size hosts for guest VMs plus CVM overhead.

AOS and hypervisor version compatibility is enforced by the compatibility matrix; mismatches block upgrades and cluster expansion.

After Foundation, run NCC health checks and confirm a healthy state before adding workloads; a new cluster starts with a default storage pool and container.

Prism Element manages one cluster; Prism Central is a separate VM (or scale-out set) that manages many clusters via registered Prism Element instances.

The cluster VIP, CVM IPs, and hypervisor management IPs must be unique and outside any DHCP pool used by guest VMs.

Prism Element shows cluster CPU, memory, storage, and IOPS for one cluster; Prism Central adds multi-cluster dashboards, categories, and policies.

NCC (Nutanix Cluster Check) runs a library of health checks; schedule it and triage failures - a failed check does not automatically mean an outage.

Alerts have severity levels (Critical, Warning, Info) and can route to email, SNMP, syslog, or webhooks; they can be acknowledged or resolved in Prism.

NTP drift is a common cause of cluster-wide issues; verify time sync across CVMs and hypervisors.

Pulse (formerly CallHome) sends telemetry to Nutanix support; disabling it limits proactive support and automated health insights.

LCM (Life Cycle Manager) handles AOS, hypervisor, firmware, and NCC updates with pre-checks and staged upgrades.

Cluster capacity is reported as usable vs. raw; RF2 consumes roughly 2x the data footprint plus overhead, so plan capacity with RF in mind.

A cluster can be registered to only one Prism Central at a time; PC must reach each PE VIP over the network.

The default bridge (br0) on each host carries CVM and hypervisor management traffic; guest VM traffic uses a separate bridge or VLAN-backed network.

AHV networks are VLAN-backed virtual networks; a VM NIC attaches to a network, and the VLAN ID defines the broadcast domain.

Bond modes: active-backup (default, no switch config), balance-slb (no LACP, source MAC hashing), and balance-tcp (requires LACP on the switch).

balance-tcp requires a correctly configured LACP port-channel upstream; misconfiguration causes connectivity loss and is a common deployment error.

The cluster VIP and CVM IPs belong on a management VLAN; isolate guest VM networks from management traffic using separate VLANs.

AHV uses Open vSwitch (OVS); Flow microsegmentation policies are defined in Prism Central using categories.

Diagnose with Prism network visualization, CVM ping/arp tests, and NCC network checks; failed CVM-to-CVM ping indicates a bridge or VLAN issue.

On ESXi clusters, the CVM uses a vSwitch port group; enable promiscuous mode and forged transmits on the dedicated port group.

A storage pool groups physical disks (HDD/SSD/NVMe); a container is a logical namespace (datastore) presented to the hypervisor, and multiple containers can share a pool.

Replication factor (RF) 2 tolerates one node/disk failure; RF3 tolerates two. RF is set per container or as a cluster default and consumes capacity accordingly.

Erasure coding (EC-X) reduces capacity overhead for cold data; it is configured per container and requires a minimum number of nodes (typically 4+).

The distributed storage fabric (DSF) is log-structured and extent-based; writes land in the local CVM staging area, then replicate to peers before acknowledgment.

Deduplication (capacity and/or inline) and compression are container-level settings; compression is inline by default on supported AOS versions.

Snapshots are metadata-based and space-efficient; they are not full copies, but changed blocks consume space over time.

Storage QoS (IOPS limits) can be applied per VM or per container to prevent noisy-neighbor effects.

Check storage health via the Prism storage dashboard, NCC disk checks, and ncli; a disk in a bad state triggers a rebuild that consumes cluster resources.

In AHV, VMs are created from images (ISOs, disk images) stored in a container; images can be uploaded via Prism, ncli, or the API.

Categories (key-value pairs) drive policy application in Prism Central for protection, recovery, and security; use categories to group VMs.

AHV supports live migration, HA (restart on host failure), and affinity/anti-affinity rules enforced by the scheduler.

VM affinity rules keep VMs on the same host; anti-affinity rules spread them across hosts. Rules can be 'should' or 'must' strength.

Prism Central Projects provide self-service VM provisioning with quotas, role-based access, and environment isolation.

Nutanix Guest Tools (NGT) enables file-level restore, application-consistent snapshots, and self-service restore for supported guest OSes.

VM power operations, console access, and snapshot/clone actions are available in Prism Element; bulk operations and policies are in Prism Central.

A VM clone can be full or linked (dependent on the parent snapshot); linked clones save space but depend on the parent image/snapshot chain.

Protection domains group VMs (by category) for replication and snapshot policies; a protection policy defines RPO, retention, and replication target.

Async replication sends snapshots to a remote cluster at the RPO interval; sync replication (Metro Availability) requires low-latency links and provides near-zero RPO.

Nutanix DR uses protection policies in Prism Central; a recovery plan defines boot order and network mapping for failover.

Snapshots are local and fast; replication is remote. A protection policy can take local snapshots only, or local plus remote replication.

NearSync replication provides RPO in seconds (typically 1-5 minutes) and requires a compatible AOS version and network latency budget.

For file-level restore, NGT must be installed in the guest; self-service restore lets users recover their own files from a snapshot.

Test failover (planned) is non-disruptive; unplanned failover promotes the remote copy. Validate the recovery plan before relying on it.

Retention policies define how many snapshots/replicas to keep (e.g., hourly/daily/weekly); retention applies per protection policy and consumes capacity on both sites.

NCC is the first diagnostic step: run it from the CVM CLI or via Prism; failed checks often point to the root cause (time sync, disk, network).

CVM memory pressure is a common cause of cluster instability; monitor CVM memory and resize if consistently high.

A disk failure triggers a rebuild; monitor rebuild progress in Prism and avoid maintenance operations during a rebuild.

NTP drift causes authentication, replication, and cluster services failures; verify NTP on CVMs and hypervisors with ntpq -p or chronyc sources.

Log collection: generate a log bundle via ncli or Prism; logs are also on the CVM under /home/nutanix/data/logs.

LCM pre-checks must pass before upgrades; common blockers are insufficient disk space, NCC failures, and version incompatibility.

A node in maintenance mode evacuates VMs and data; use it before hardware service to avoid data loss and keep RF satisfied.

Cluster health is not just green lights - check the health dashboard, NCC, and capacity trends; a cluster can be functional but degraded (e.g., RF not satisfied).

Prism Central is the multicloud control plane; it manages multiple Prism Element clusters (on-prem and in public cloud) from one UI.

Nutanix Clusters on AWS/Azure extends AOS networking and management to public cloud; it is not the same as running AHV on-prem.

Calm (Nutanix Cloud Manager) provides orchestration and self-service for multi-cloud workloads; blueprints define multi-tier applications.

Nutanix Cloud Manager (NCM) includes cost governance, security, and self-service; it integrates with Prism Central and public cloud accounts.

Flow (microsegmentation) policies are defined in Prism Central and applied across clusters; they use categories, not IP addresses, for policy targets.

Multi-cluster management requires network connectivity from Prism Central to each PE VIP and consistent time sync; a cluster can be registered to only one PC.

Categories are the key mechanism for applying policies (protection, security, cost) consistently across clusters and clouds.

Cloud-native workloads can be managed alongside AHV VMs; Nutanix provides CSI/CNI integrations for Kubernetes (NKP) in multicloud scenarios.

Foundation discovers nodes via IPMI/Redfish out-of-band; it wipes disks, installs the hypervisor and AOS, and creates the cluster from a jump host.

Minimum 3 nodes for RF2, 4 nodes for RF3; single-node clusters exist but provide no redundancy.

Cluster creation requires cluster name, VIP, CVM IPs, hypervisor IPs, subnet mask, gateway, DNS, and NTP; the VIP must be a free IP in the CVM subnet.

Default CVM memory is 16 GB (32 GB for large nodes); the CVM reserves host memory, so size hosts for guest VMs plus CVM overhead.

AOS and hypervisor version compatibility is enforced by the compatibility matrix; mismatches block upgrades and cluster expansion.

After Foundation, run NCC health checks and confirm a healthy state before adding workloads; a new cluster starts with a default storage pool and container.

Prism Element manages one cluster; Prism Central is a separate VM (or scale-out set) that manages many clusters via registered Prism Element instances.

The cluster VIP, CVM IPs, and hypervisor management IPs must be unique and outside any DHCP pool used by guest VMs.

Prism Element shows cluster CPU, memory, storage, and IOPS for one cluster; Prism Central adds multi-cluster dashboards, categories, and policies.

NCC (Nutanix Cluster Check) runs a library of health checks; schedule it and triage failures - a failed check does not automatically mean an outage.

Alerts have severity levels (Critical, Warning, Info) and can route to email, SNMP, syslog, or webhooks; they can be acknowledged or resolved in Prism.

NTP drift is a common cause of cluster-wide issues; verify time sync across CVMs and hypervisors.

Pulse (formerly CallHome) sends telemetry to Nutanix support; disabling it limits proactive support and automated health insights.

LCM (Life Cycle Manager) handles AOS, hypervisor, firmware, and NCC updates with pre-checks and staged upgrades.

Cluster capacity is reported as usable vs. raw; RF2 consumes roughly 2x the data footprint plus overhead, so plan capacity with RF in mind.

A cluster can be registered to only one Prism Central at a time; PC must reach each PE VIP over the network.

The default bridge (br0) on each host carries CVM and hypervisor management traffic; guest VM traffic uses a separate bridge or VLAN-backed network.

AHV networks are VLAN-backed virtual networks; a VM NIC attaches to a network, and the VLAN ID defines the broadcast domain.

Bond modes: active-backup (default, no switch config), balance-slb (no LACP, source MAC hashing), and balance-tcp (requires LACP on the switch).

balance-tcp requires a correctly configured LACP port-channel upstream; misconfiguration causes connectivity loss and is a common deployment error.

The cluster VIP and CVM IPs belong on a management VLAN; isolate guest VM networks from management traffic using separate VLANs.

AHV uses Open vSwitch (OVS); Flow microsegmentation policies are defined in Prism Central using categories.

Diagnose with Prism network visualization, CVM ping/arp tests, and NCC network checks; failed CVM-to-CVM ping indicates a bridge or VLAN issue.

On ESXi clusters, the CVM uses a vSwitch port group; enable promiscuous mode and forged transmits on the dedicated port group.

A storage pool groups physical disks (HDD/SSD/NVMe); a container is a logical namespace (datastore) presented to the hypervisor, and multiple containers can share a pool.

Replication factor (RF) 2 tolerates one node/disk failure; RF3 tolerates two. RF is set per container or as a cluster default and consumes capacity accordingly.

Erasure coding (EC-X) reduces capacity overhead for cold data; it is configured per container and requires a minimum number of nodes (typically 4+).

The distributed storage fabric (DSF) is log-structured and extent-based; writes land in the local CVM staging area, then replicate to peers before acknowledgment.

Deduplication (capacity and/or inline) and compression are container-level settings; compression is inline by default on supported AOS versions.

Snapshots are metadata-based and space-efficient; they are not full copies, but changed blocks consume space over time.

Storage QoS (IOPS limits) can be applied per VM or per container to prevent noisy-neighbor effects.

Check storage health via the Prism storage dashboard, NCC disk checks, and ncli; a disk in a bad state triggers a rebuild that consumes cluster resources.

In AHV, VMs are created from images (ISOs, disk images) stored in a container; images can be uploaded via Prism, ncli, or the API.

Categories (key-value pairs) drive policy application in Prism Central for protection, recovery, and security; use categories to group VMs.

AHV supports live migration, HA (restart on host failure), and affinity/anti-affinity rules enforced by the scheduler.

VM affinity rules keep VMs on the same host; anti-affinity rules spread them across hosts. Rules can be 'should' or 'must' strength.

Prism Central Projects provide self-service VM provisioning with quotas, role-based access, and environment isolation.

Nutanix Guest Tools (NGT) enables file-level restore, application-consistent snapshots, and self-service restore for supported guest OSes.

VM power operations, console access, and snapshot/clone actions are available in Prism Element; bulk operations and policies are in Prism Central.

A VM clone can be full or linked (dependent on the parent snapshot); linked clones save space but depend on the parent image/snapshot chain.

Protection domains group VMs (by category) for replication and snapshot policies; a protection policy defines RPO, retention, and replication target.

Async replication sends snapshots to a remote cluster at the RPO interval; sync replication (Metro Availability) requires low-latency links and provides near-zero RPO.

Nutanix DR uses protection policies in Prism Central; a recovery plan defines boot order and network mapping for failover.

Snapshots are local and fast; replication is remote. A protection policy can take local snapshots only, or local plus remote replication.

NearSync replication provides RPO in seconds (typically 1-5 minutes) and requires a compatible AOS version and network latency budget.

For file-level restore, NGT must be installed in the guest; self-service restore lets users recover their own files from a snapshot.

Test failover (planned) is non-disruptive; unplanned failover promotes the remote copy. Validate the recovery plan before relying on it.

Retention policies define how many snapshots/replicas to keep (e.g., hourly/daily/weekly); retention applies per protection policy and consumes capacity on both sites.

NCC is the first diagnostic step: run it from the CVM CLI or via Prism; failed checks often point to the root cause (time sync, disk, network).

CVM memory pressure is a common cause of cluster instability; monitor CVM memory and resize if consistently high.

A disk failure triggers a rebuild; monitor rebuild progress in Prism and avoid maintenance operations during a rebuild.

NTP drift causes authentication, replication, and cluster services failures; verify NTP on CVMs and hypervisors with ntpq -p or chronyc sources.

Log collection: generate a log bundle via ncli or Prism; logs are also on the CVM under /home/nutanix/data/logs.

LCM pre-checks must pass before upgrades; common blockers are insufficient disk space, NCC failures, and version incompatibility.

A node in maintenance mode evacuates VMs and data; use it before hardware service to avoid data loss and keep RF satisfied.

Cluster health is not just green lights - check the health dashboard, NCC, and capacity trends; a cluster can be functional but degraded (e.g., RF not satisfied).

Prism Central is the multicloud control plane; it manages multiple Prism Element clusters (on-prem and in public cloud) from one UI.

Nutanix Clusters on AWS/Azure extends AOS networking and management to public cloud; it is not the same as running AHV on-prem.

Calm (Nutanix Cloud Manager) provides orchestration and self-service for multi-cloud workloads; blueprints define multi-tier applications.

Nutanix Cloud Manager (NCM) includes cost governance, security, and self-service; it integrates with Prism Central and public cloud accounts.

Flow (microsegmentation) policies are defined in Prism Central and applied across clusters; they use categories, not IP addresses, for policy targets.

Multi-cluster management requires network connectivity from Prism Central to each PE VIP and consistent time sync; a cluster can be registered to only one PC.

Categories are the key mechanism for applying policies (protection, security, cost) consistently across clusters and clouds.

Cloud-native workloads can be managed alongside AHV VMs; Nutanix provides CSI/CNI integrations for Kubernetes (NKP) in multicloud scenarios.

Foundation discovers nodes via IPMI/Redfish out-of-band; it wipes disks, installs the hypervisor and AOS, and creates the cluster from a jump host.

Minimum 3 nodes for RF2, 4 nodes for RF3; single-node clusters exist but provide no redundancy.

Cluster creation requires cluster name, VIP, CVM IPs, hypervisor IPs, subnet mask, gateway, DNS, and NTP; the VIP must be a free IP in the CVM subnet.

Default CVM memory is 16 GB (32 GB for large nodes); the CVM reserves host memory, so size hosts for guest VMs plus CVM overhead.

AOS and hypervisor version compatibility is enforced by the compatibility matrix; mismatches block upgrades and cluster expansion.

After Foundation, run NCC health checks and confirm a healthy state before adding workloads; a new cluster starts with a default storage pool and container.

Prism Element manages one cluster; Prism Central is a separate VM (or scale-out set) that manages many clusters via registered Prism Element instances.

The cluster VIP, CVM IPs, and hypervisor management IPs must be unique and outside any DHCP pool used by guest VMs.

Prism Element shows cluster CPU, memory, storage, and IOPS for one cluster; Prism Central adds multi-cluster dashboards, categories, and policies.

NCC (Nutanix Cluster Check) runs a library of health checks; schedule it and triage failures - a failed check does not automatically mean an outage.

Alerts have severity levels (Critical, Warning, Info) and can route to email, SNMP, syslog, or webhooks; they can be acknowledged or resolved in Prism.

NTP drift is a common cause of cluster-wide issues; verify time sync across CVMs and hypervisors.

Pulse (formerly CallHome) sends telemetry to Nutanix support; disabling it limits proactive support and automated health insights.

LCM (Life Cycle Manager) handles AOS, hypervisor, firmware, and NCC updates with pre-checks and staged upgrades.

Cluster capacity is reported as usable vs. raw; RF2 consumes roughly 2x the data footprint plus overhead, so plan capacity with RF in mind.

A cluster can be registered to only one Prism Central at a time; PC must reach each PE VIP over the network.

The default bridge (br0) on each host carries CVM and hypervisor management traffic; guest VM traffic uses a separate bridge or VLAN-backed network.

AHV networks are VLAN-backed virtual networks; a VM NIC attaches to a network, and the VLAN ID defines the broadcast domain.

Bond modes: active-backup (default, no switch config), balance-slb (no LACP, source MAC hashing), and balance-tcp (requires LACP on the switch).

balance-tcp requires a correctly configured LACP port-channel upstream; misconfiguration causes connectivity loss and is a common deployment error.

The cluster VIP and CVM IPs belong on a management VLAN; isolate guest VM networks from management traffic using separate VLANs.

AHV uses Open vSwitch (OVS); Flow microsegmentation policies are defined in Prism Central using categories.

Diagnose with Prism network visualization, CVM ping/arp tests, and NCC network checks; failed CVM-to-CVM ping indicates a bridge or VLAN issue.

On ESXi clusters, the CVM uses a vSwitch port group; enable promiscuous mode and forged transmits on the dedicated port group.

A storage pool groups physical disks (HDD/SSD/NVMe); a container is a logical namespace (datastore) presented to the hypervisor, and multiple containers can share a pool.

Replication factor (RF) 2 tolerates one node/disk failure; RF3 tolerates two. RF is set per container or as a cluster default and consumes capacity accordingly.

Erasure coding (EC-X) reduces capacity overhead for cold data; it is configured per container and requires a minimum number of nodes (typically 4+).

The distributed storage fabric (DSF) is log-structured and extent-based; writes land in the local CVM staging area, then replicate to peers before acknowledgment.

Deduplication (capacity and/or inline) and compression are container-level settings; compression is inline by default on supported AOS versions.

Snapshots are metadata-based and space-efficient; they are not full copies, but changed blocks consume space over time.

Storage QoS (IOPS limits) can be applied per VM or per container to prevent noisy-neighbor effects.

Check storage health via the Prism storage dashboard, NCC disk checks, and ncli; a disk in a bad state triggers a rebuild that consumes cluster resources.

In AHV, VMs are created from images (ISOs, disk images) stored in a container; images can be uploaded via Prism, ncli, or the API.

Categories (key-value pairs) drive policy application in Prism Central for protection, recovery, and security; use categories to group VMs.

AHV supports live migration, HA (restart on host failure), and affinity/anti-affinity rules enforced by the scheduler.

VM affinity rules keep VMs on the same host; anti-affinity rules spread them across hosts. Rules can be 'should' or 'must' strength.

Prism Central Projects provide self-service VM provisioning with quotas, role-based access, and environment isolation.

Nutanix Guest Tools (NGT) enables file-level restore, application-consistent snapshots, and self-service restore for supported guest OSes.

VM power operations, console access, and snapshot/clone actions are available in Prism Element; bulk operations and policies are in Prism Central.

A VM clone can be full or linked (dependent on the parent snapshot); linked clones save space but depend on the parent image/snapshot chain.

Protection domains group VMs (by category) for replication and snapshot policies; a protection policy defines RPO, retention, and replication target.

Async replication sends snapshots to a remote cluster at the RPO interval; sync replication (Metro Availability) requires low-latency links and provides near-zero RPO.

Nutanix DR uses protection policies in Prism Central; a recovery plan defines boot order and network mapping for failover.

Snapshots are local and fast; replication is remote. A protection policy can take local snapshots only, or local plus remote replication.

NearSync replication provides RPO in seconds (typically 1-5 minutes) and requires a compatible AOS version and network latency budget.

For file-level restore, NGT must be installed in the guest; self-service restore lets users recover their own files from a snapshot.

Test failover (planned) is non-disruptive; unplanned failover promotes the remote copy. Validate the recovery plan before relying on it.

Retention policies define how many snapshots/replicas to keep (e.g., hourly/daily/weekly); retention applies per protection policy and consumes capacity on both sites.

NCC is the first diagnostic step: run it from the CVM CLI or via Prism; failed checks often point to the root cause (time sync, disk, network).

CVM memory pressure is a common cause of cluster instability; monitor CVM memory and resize if consistently high.

A disk failure triggers a rebuild; monitor rebuild progress in Prism and avoid maintenance operations during a rebuild.

NTP drift causes authentication, replication, and cluster services failures; verify NTP on CVMs and hypervisors with ntpq -p or chronyc sources.

Log collection: generate a log bundle via ncli or Prism; logs are also on the CVM under /home/nutanix/data/logs.

LCM pre-checks must pass before upgrades; common blockers are insufficient disk space, NCC failures, and version incompatibility.

A node in maintenance mode evacuates VMs and data; use it before hardware service to avoid data loss and keep RF satisfied.

Cluster health is not just green lights - check the health dashboard, NCC, and capacity trends; a cluster can be functional but degraded (e.g., RF not satisfied).

Prism Central is the multicloud control plane; it manages multiple Prism Element clusters (on-prem and in public cloud) from one UI.

Nutanix Clusters on AWS/Azure extends AOS networking and management to public cloud; it is not the same as running AHV on-prem.

Calm (Nutanix Cloud Manager) provides orchestration and self-service for multi-cloud workloads; blueprints define multi-tier applications.

Nutanix Cloud Manager (NCM) includes cost governance, security, and self-service; it integrates with Prism Central and public cloud accounts.

Flow (microsegmentation) policies are defined in Prism Central and applied across clusters; they use categories, not IP addresses, for policy targets.

Multi-cluster management requires network connectivity from Prism Central to each PE VIP and consistent time sync; a cluster can be registered to only one PC.

Categories are the key mechanism for applying policies (protection, security, cost) consistently across clusters and clouds.

Cloud-native workloads can be managed alongside AHV VMs; Nutanix provides CSI/CNI integrations for Kubernetes (NKP) in multicloud scenarios.

Foundation discovers nodes via IPMI/Redfish out-of-band; it wipes disks, installs the hypervisor and AOS, and creates the cluster from a jump host.

Minimum 3 nodes for RF2, 4 nodes for RF3; single-node clusters exist but provide no redundancy.

Cluster creation requires cluster name, VIP, CVM IPs, hypervisor IPs, subnet mask, gateway, DNS, and NTP; the VIP must be a free IP in the CVM subnet.

Default CVM memory is 16 GB (32 GB for large nodes); the CVM reserves host memory, so size hosts for guest VMs plus CVM overhead.

AOS and hypervisor version compatibility is enforced by the compatibility matrix; mismatches block upgrades and cluster expansion.

After Foundation, run NCC health checks and confirm a healthy state before adding workloads; a new cluster starts with a default storage pool and container.

Prism Element manages one cluster; Prism Central is a separate VM (or scale-out set) that manages many clusters via registered Prism Element instances.

The cluster VIP, CVM IPs, and hypervisor management IPs must be unique and outside any DHCP pool used by guest VMs.

Prism Element shows cluster CPU, memory, storage, and IOPS for one cluster; Prism Central adds multi-cluster dashboards, categories, and policies.

NCC (Nutanix Cluster Check) runs a library of health checks; schedule it and triage failures - a failed check does not automatically mean an outage.

Alerts have severity levels (Critical, Warning, Info) and can route to email, SNMP, syslog, or webhooks; they can be acknowledged or resolved in Prism.

NTP drift is a common cause of cluster-wide issues; verify time sync across CVMs and hypervisors.

Pulse (formerly CallHome) sends telemetry to Nutanix support; disabling it limits proactive support and automated health insights.

LCM (Life Cycle Manager) handles AOS, hypervisor, firmware, and NCC updates with pre-checks and staged upgrades.

Cluster capacity is reported as usable vs. raw; RF2 consumes roughly 2x the data footprint plus overhead, so plan capacity with RF in mind.

A cluster can be registered to only one Prism Central at a time; PC must reach each PE VIP over the network.

The default bridge (br0) on each host carries CVM and hypervisor management traffic; guest VM traffic uses a separate bridge or VLAN-backed network.

AHV networks are VLAN-backed virtual networks; a VM NIC attaches to a network, and the VLAN ID defines the broadcast domain.

Bond modes: active-backup (default, no switch config), balance-slb (no LACP, source MAC hashing), and balance-tcp (requires LACP on the switch).

balance-tcp requires a correctly configured LACP port-channel upstream; misconfiguration causes connectivity loss and is a common deployment error.

The cluster VIP and CVM IPs belong on a management VLAN; isolate guest VM networks from management traffic using separate VLANs.

AHV uses Open vSwitch (OVS); Flow microsegmentation policies are defined in Prism Central using categories.

Diagnose with Prism network visualization, CVM ping/arp tests, and NCC network checks; failed CVM-to-CVM ping indicates a bridge or VLAN issue.

On ESXi clusters, the CVM uses a vSwitch port group; enable promiscuous mode and forged transmits on the dedicated port group.

A storage pool groups physical disks (HDD/SSD/NVMe); a container is a logical namespace (datastore) presented to the hypervisor, and multiple containers can share a pool.

Replication factor (RF) 2 tolerates one node/disk failure; RF3 tolerates two. RF is set per container or as a cluster default and consumes capacity accordingly.

Erasure coding (EC-X) reduces capacity overhead for cold data; it is configured per container and requires a minimum number of nodes (typically 4+).

The distributed storage fabric (DSF) is log-structured and extent-based; writes land in the local CVM staging area, then replicate to peers before acknowledgment.

Deduplication (capacity and/or inline) and compression are container-level settings; compression is inline by default on supported AOS versions.

Snapshots are metadata-based and space-efficient; they are not full copies, but changed blocks consume space over time.

Storage QoS (IOPS limits) can be applied per VM or per container to prevent noisy-neighbor effects.

Check storage health via the Prism storage dashboard, NCC disk checks, and ncli; a disk in a bad state triggers a rebuild that consumes cluster resources.

In AHV, VMs are created from images (ISOs, disk images) stored in a container; images can be uploaded via Prism, ncli, or the API.

Categories (key-value pairs) drive policy application in Prism Central for protection, recovery, and security; use categories to group VMs.

AHV supports live migration, HA (restart on host failure), and affinity/anti-affinity rules enforced by the scheduler.

VM affinity rules keep VMs on the same host; anti-affinity rules spread them across hosts. Rules can be 'should' or 'must' strength.

Prism Central Projects provide self-service VM provisioning with quotas, role-based access, and environment isolation.

Nutanix Guest Tools (NGT) enables file-level restore, application-consistent snapshots, and self-service restore for supported guest OSes.

VM power operations, console access, and snapshot/clone actions are available in Prism Element; bulk operations and policies are in Prism Central.

A VM clone can be full or linked (dependent on the parent snapshot); linked clones save space but depend on the parent image/snapshot chain.

Protection domains group VMs (by category) for replication and snapshot policies; a protection policy defines RPO, retention, and replication target.

Async replication sends snapshots to a remote cluster at the RPO interval; sync replication (Metro Availability) requires low-latency links and provides near-zero RPO.

Nutanix DR uses protection policies in Prism Central; a recovery plan defines boot order and network mapping for failover.

Snapshots are local and fast; replication is remote. A protection policy can take local snapshots only, or local plus remote replication.

NearSync replication provides RPO in seconds (typically 1-5 minutes) and requires a compatible AOS version and network latency budget.

For file-level restore, NGT must be installed in the guest; self-service restore lets users recover their own files from a snapshot.

Test failover (planned) is non-disruptive; unplanned failover promotes the remote copy. Validate the recovery plan before relying on it.

Retention policies define how many snapshots/replicas to keep (e.g., hourly/daily/weekly); retention applies per protection policy and consumes capacity on both sites.

NCC is the first diagnostic step: run it from the CVM CLI or via Prism; failed checks often point to the root cause (time sync, disk, network).

CVM memory pressure is a common cause of cluster instability; monitor CVM memory and resize if consistently high.

A disk failure triggers a rebuild; monitor rebuild progress in Prism and avoid maintenance operations during a rebuild.

NTP drift causes authentication, replication, and cluster services failures; verify NTP on CVMs and hypervisors with ntpq -p or chronyc sources.

Log collection: generate a log bundle via ncli or Prism; logs are also on the CVM under /home/nutanix/data/logs.

LCM pre-checks must pass before upgrades; common blockers are insufficient disk space, NCC failures, and version incompatibility.

A node in maintenance mode evacuates VMs and data; use it before hardware service to avoid data loss and keep RF satisfied.

Cluster health is not just green lights - check the health dashboard, NCC, and capacity trends; a cluster can be functional but degraded (e.g., RF not satisfied).

Prism Central is the multicloud control plane; it manages multiple Prism Element clusters (on-prem and in public cloud) from one UI.

Nutanix Clusters on AWS/Azure extends AOS networking and management to public cloud; it is not the same as running AHV on-prem.

Calm (Nutanix Cloud Manager) provides orchestration and self-service for multi-cloud workloads; blueprints define multi-tier applications.

Nutanix Cloud Manager (NCM) includes cost governance, security, and self-service; it integrates with Prism Central and public cloud accounts.

Flow (microsegmentation) policies are defined in Prism Central and applied across clusters; they use categories, not IP addresses, for policy targets.

Multi-cluster management requires network connectivity from Prism Central to each PE VIP and consistent time sync; a cluster can be registered to only one PC.

Categories are the key mechanism for applying policies (protection, security, cost) consistently across clusters and clouds.

Cloud-native workloads can be managed alongside AHV VMs; Nutanix provides CSI/CNI integrations for Kubernetes (NKP) in multicloud scenarios.

Foundation discovers nodes via IPMI/Redfish out-of-band; it wipes disks, installs the hypervisor and AOS, and creates the cluster from a jump host.

Minimum 3 nodes for RF2, 4 nodes for RF3; single-node clusters exist but provide no redundancy.

Cluster creation requires cluster name, VIP, CVM IPs, hypervisor IPs, subnet mask, gateway, DNS, and NTP; the VIP must be a free IP in the CVM subnet.

Default CVM memory is 16 GB (32 GB for large nodes); the CVM reserves host memory, so size hosts for guest VMs plus CVM overhead.

AOS and hypervisor version compatibility is enforced by the compatibility matrix; mismatches block upgrades and cluster expansion.

After Foundation, run NCC health checks and confirm a healthy state before adding workloads; a new cluster starts with a default storage pool and container.

Prism Element manages one cluster; Prism Central is a separate VM (or scale-out set) that manages many clusters via registered Prism Element instances.

The cluster VIP, CVM IPs, and hypervisor management IPs must be unique and outside any DHCP pool used by guest VMs.

Prism Element shows cluster CPU, memory, storage, and IOPS for one cluster; Prism Central adds multi-cluster dashboards, categories, and policies.

NCC (Nutanix Cluster Check) runs a library of health checks; schedule it and triage failures - a failed check does not automatically mean an outage.

Alerts have severity levels (Critical, Warning, Info) and can route to email, SNMP, syslog, or webhooks; they can be acknowledged or resolved in Prism.

NTP drift is a common cause of cluster-wide issues; verify time sync across CVMs and hypervisors.

Pulse (formerly CallHome) sends telemetry to Nutanix support; disabling it limits proactive support and automated health insights.

LCM (Life Cycle Manager) handles AOS, hypervisor, firmware, and NCC updates with pre-checks and staged upgrades.

Cluster capacity is reported as usable vs. raw; RF2 consumes roughly 2x the data footprint plus overhead, so plan capacity with RF in mind.

A cluster can be registered to only one Prism Central at a time; PC must reach each PE VIP over the network.

The default bridge (br0) on each host carries CVM and hypervisor management traffic; guest VM traffic uses a separate bridge or VLAN-backed network.

AHV networks are VLAN-backed virtual networks; a VM NIC attaches to a network, and the VLAN ID defines the broadcast domain.

Bond modes: active-backup (default, no switch config), balance-slb (no LACP, source MAC hashing), and balance-tcp (requires LACP on the switch).

balance-tcp requires a correctly configured LACP port-channel upstream; misconfiguration causes connectivity loss and is a common deployment error.

The cluster VIP and CVM IPs belong on a management VLAN; isolate guest VM networks from management traffic using separate VLANs.

AHV uses Open vSwitch (OVS); Flow microsegmentation policies are defined in Prism Central using categories.

Diagnose with Prism network visualization, CVM ping/arp tests, and NCC network checks; failed CVM-to-CVM ping indicates a bridge or VLAN issue.

On ESXi clusters, the CVM uses a vSwitch port group; enable promiscuous mode and forged transmits on the dedicated port group.

A storage pool groups physical disks (HDD/SSD/NVMe); a container is a logical namespace (datastore) presented to the hypervisor, and multiple containers can share a pool.

Replication factor (RF) 2 tolerates one node/disk failure; RF3 tolerates two. RF is set per container or as a cluster default and consumes capacity accordingly.

Erasure coding (EC-X) reduces capacity overhead for cold data; it is configured per container and requires a minimum number of nodes (typically 4+).

The distributed storage fabric (DSF) is log-structured and extent-based; writes land in the local CVM staging area, then replicate to peers before acknowledgment.

Deduplication (capacity and/or inline) and compression are container-level settings; compression is inline by default on supported AOS versions.

Snapshots are metadata-based and space-efficient; they are not full copies, but changed blocks consume space over time.

Storage QoS (IOPS limits) can be applied per VM or per container to prevent noisy-neighbor effects.

Check storage health via the Prism storage dashboard, NCC disk checks, and ncli; a disk in a bad state triggers a rebuild that consumes cluster resources.

In AHV, VMs are created from images (ISOs, disk images) stored in a container; images can be uploaded via Prism, ncli, or the API.

Categories (key-value pairs) drive policy application in Prism Central for protection, recovery, and security; use categories to group VMs.

AHV supports live migration, HA (restart on host failure), and affinity/anti-affinity rules enforced by the scheduler.

VM affinity rules keep VMs on the same host; anti-affinity rules spread them across hosts. Rules can be 'should' or 'must' strength.

Prism Central Projects provide self-service VM provisioning with quotas, role-based access, and environment isolation.

Nutanix Guest Tools (NGT) enables file-level restore, application-consistent snapshots, and self-service restore for supported guest OSes.

VM power operations, console access, and snapshot/clone actions are available in Prism Element; bulk operations and policies are in Prism Central.

A VM clone can be full or linked (dependent on the parent snapshot); linked clones save space but depend on the parent image/snapshot chain.

Protection domains group VMs (by category) for replication and snapshot policies; a protection policy defines RPO, retention, and replication target.

Async replication sends snapshots to a remote cluster at the RPO interval; sync replication (Metro Availability) requires low-latency links and provides near-zero RPO.

Nutanix DR uses protection policies in Prism Central; a recovery plan defines boot order and network mapping for failover.

Snapshots are local and fast; replication is remote. A protection policy can take local snapshots only, or local plus remote replication.

NearSync replication provides RPO in seconds (typically 1-5 minutes) and requires a compatible AOS version and network latency budget.

For file-level restore, NGT must be installed in the guest; self-service restore lets users recover their own files from a snapshot.

Test failover (planned) is non-disruptive; unplanned failover promotes the remote copy. Validate the recovery plan before relying on it.

Retention policies define how many snapshots/replicas to keep (e.g., hourly/daily/weekly); retention applies per protection policy and consumes capacity on both sites.

NCC is the first diagnostic step: run it from the CVM CLI or via Prism; failed checks often point to the root cause (time sync, disk, network).

CVM memory pressure is a common cause of cluster instability; monitor CVM memory and resize if consistently high.

A disk failure triggers a rebuild; monitor rebuild progress in Prism and avoid maintenance operations during a rebuild.

NTP drift causes authentication, replication, and cluster services failures; verify NTP on CVMs and hypervisors with ntpq -p or chronyc sources.

Log collection: generate a log bundle via ncli or Prism; logs are also on the CVM under /home/nutanix/data/logs.

LCM pre-checks must pass before upgrades; common blockers are insufficient disk space, NCC failures, and version incompatibility.

A node in maintenance mode evacuates VMs and data; use it before hardware service to avoid data loss and keep RF satisfied.

Cluster health is not just green lights - check the health dashboard, NCC, and capacity trends; a cluster can be functional but degraded (e.g., RF not satisfied).

Prism Central is the multicloud control plane; it manages multiple Prism Element clusters (on-prem and in public cloud) from one UI.

Nutanix Clusters on AWS/Azure extends AOS networking and management to public cloud; it is not the same as running AHV on-prem.

Calm (Nutanix Cloud Manager) provides orchestration and self-service for multi-cloud workloads; blueprints define multi-tier applications.

Nutanix Cloud Manager (NCM) includes cost governance, security, and self-service; it integrates with Prism Central and public cloud accounts.

Flow (microsegmentation) policies are defined in Prism Central and applied across clusters; they use categories, not IP addresses, for policy targets.

Multi-cluster management requires network connectivity from Prism Central to each PE VIP and consistent time sync; a cluster can be registered to only one PC.

Categories are the key mechanism for applying policies (protection, security, cost) consistently across clusters and clouds.

Cloud-native workloads can be managed alongside AHV VMs; Nutanix provides CSI/CNI integrations for Kubernetes (NKP) in multicloud scenarios.

Foundation discovers nodes via IPMI/Redfish out-of-band; it wipes disks, installs the hypervisor and AOS, and creates the cluster from a jump host.

Minimum 3 nodes for RF2, 4 nodes for RF3; single-node clusters exist but provide no redundancy.

Cluster creation requires cluster name, VIP, CVM IPs, hypervisor IPs, subnet mask, gateway, DNS, and NTP; the VIP must be a free IP in the CVM subnet.

Default CVM memory is 16 GB (32 GB for large nodes); the CVM reserves host memory, so size hosts for guest VMs plus CVM overhead.

AOS and hypervisor version compatibility is enforced by the compatibility matrix; mismatches block upgrades and cluster expansion.

After Foundation, run NCC health checks and confirm a healthy state before adding workloads; a new cluster starts with a default storage pool and container.

Prism Element manages one cluster; Prism Central is a separate VM (or scale-out set) that manages many clusters via registered Prism Element instances.

The cluster VIP, CVM IPs, and hypervisor management IPs must be unique and outside any DHCP pool used by guest VMs.

Prism Element shows cluster CPU, memory, storage, and IOPS for one cluster; Prism Central adds multi-cluster dashboards, categories, and policies.

NCC (Nutanix Cluster Check) runs a library of health checks; schedule it and triage failures - a failed check does not automatically mean an outage.

Alerts have severity levels (Critical, Warning, Info) and can route to email, SNMP, syslog, or webhooks; they can be acknowledged or resolved in Prism.

NTP drift is a common cause of cluster-wide issues; verify time sync across CVMs and hypervisors.

Pulse (formerly CallHome) sends telemetry to Nutanix support; disabling it limits proactive support and automated health insights.

LCM (Life Cycle Manager) handles AOS, hypervisor, firmware, and NCC updates with pre-checks and staged upgrades.

Cluster capacity is reported as usable vs. raw; RF2 consumes roughly 2x the data footprint plus overhead, so plan capacity with RF in mind.

A cluster can be registered to only one Prism Central at a time; PC must reach each PE VIP over the network.

The default bridge (br0) on each host carries CVM and hypervisor management traffic; guest VM traffic uses a separate bridge or VLAN-backed network.

AHV networks are VLAN-backed virtual networks; a VM NIC attaches to a network, and the VLAN ID defines the broadcast domain.

Bond modes: active-backup (default, no switch config), balance-slb (no LACP, source MAC hashing), and balance-tcp (requires LACP on the switch).

balance-tcp requires a correctly configured LACP port-channel upstream; misconfiguration causes connectivity loss and is a common deployment error.

The cluster VIP and CVM IPs belong on a management VLAN; isolate guest VM networks from management traffic using separate VLANs.

AHV uses Open vSwitch (OVS); Flow microsegmentation policies are defined in Prism Central using categories.

Diagnose with Prism network visualization, CVM ping/arp tests, and NCC network checks; failed CVM-to-CVM ping indicates a bridge or VLAN issue.

On ESXi clusters, the CVM uses a vSwitch port group; enable promiscuous mode and forged transmits on the dedicated port group.

A storage pool groups physical disks (HDD/SSD/NVMe); a container is a logical namespace (datastore) presented to the hypervisor, and multiple containers can share a pool.

Replication factor (RF) 2 tolerates one node/disk failure; RF3 tolerates two. RF is set per container or as a cluster default and consumes capacity accordingly.

Erasure coding (EC-X) reduces capacity overhead for cold data; it is configured per container and requires a minimum number of nodes (typically 4+).

The distributed storage fabric (DSF) is log-structured and extent-based; writes land in the local CVM staging area, then replicate to peers before acknowledgment.

Deduplication (capacity and/or inline) and compression are container-level settings; compression is inline by default on supported AOS versions.

Snapshots are metadata-based and space-efficient; they are not full copies, but changed blocks consume space over time.

Storage QoS (IOPS limits) can be applied per VM or per container to prevent noisy-neighbor effects.

Check storage health via the Prism storage dashboard, NCC disk checks, and ncli; a disk in a bad state triggers a rebuild that consumes cluster resources.

In AHV, VMs are created from images (ISOs, disk images) stored in a container; images can be uploaded via Prism, ncli, or the API.

Categories (key-value pairs) drive policy application in Prism Central for protection, recovery, and security; use categories to group VMs.

AHV supports live migration, HA (restart on host failure), and affinity/anti-affinity rules enforced by the scheduler.

VM affinity rules keep VMs on the same host; anti-affinity rules spread them across hosts. Rules can be 'should' or 'must' strength.

Prism Central Projects provide self-service VM provisioning with quotas, role-based access, and environment isolation.

Nutanix Guest Tools (NGT) enables file-level restore, application-consistent snapshots, and self-service restore for supported guest OSes.

VM power operations, console access, and snapshot/clone actions are available in Prism Element; bulk operations and policies are in Prism Central.

A VM clone can be full or linked (dependent on the parent snapshot); linked clones save space but depend on the parent image/snapshot chain.

Protection domains group VMs (by category) for replication and snapshot policies; a protection policy defines RPO, retention, and replication target.

Async replication sends snapshots to a remote cluster at the RPO interval; sync replication (Metro Availability) requires low-latency links and provides near-zero RPO.

Nutanix DR uses protection policies in Prism Central; a recovery plan defines boot order and network mapping for failover.

Snapshots are local and fast; replication is remote. A protection policy can take local snapshots only, or local plus remote replication.

NearSync replication provides RPO in seconds (typically 1-5 minutes) and requires a compatible AOS version and network latency budget.

For file-level restore, NGT must be installed in the guest; self-service restore lets users recover their own files from a snapshot.

Test failover (planned) is non-disruptive; unplanned failover promotes the remote copy. Validate the recovery plan before relying on it.

Retention policies define how many snapshots/replicas to keep (e.g., hourly/daily/weekly); retention applies per protection policy and consumes capacity on both sites.

NCC is the first diagnostic step: run it from the CVM CLI or via Prism; failed checks often point to the root cause (time sync, disk, network).

CVM memory pressure is a common cause of cluster instability; monitor CVM memory and resize if consistently high.

A disk failure triggers a rebuild; monitor rebuild progress in Prism and avoid maintenance operations during a rebuild.

NTP drift causes authentication, replication, and cluster services failures; verify NTP on CVMs and hypervisors with ntpq -p or chronyc sources.

Log collection: generate a log bundle via ncli or Prism; logs are also on the CVM under /home/nutanix/data/logs.

LCM pre-checks must pass before upgrades; common blockers are insufficient disk space, NCC failures, and version incompatibility.

A node in maintenance mode evacuates VMs and data; use it before hardware service to avoid data loss and keep RF satisfied.

Cluster health is not just green lights - check the health dashboard, NCC, and capacity trends; a cluster can be functional but degraded (e.g., RF not satisfied).

Prism Central is the multicloud control plane; it manages multiple Prism Element clusters (on-prem and in public cloud) from one UI.

Nutanix Clusters on AWS/Azure extends AOS networking and management to public cloud; it is not the same as running AHV on-prem.

Calm (Nutanix Cloud Manager) provides orchestration and self-service for multi-cloud workloads; blueprints define multi-tier applications.

Nutanix Cloud Manager (NCM) includes cost governance, security, and self-service; it integrates with Prism Central and public cloud accounts.

Flow (microsegmentation) policies are defined in Prism Central and applied across clusters; they use categories, not IP addresses, for policy targets.

Multi-cluster management requires network connectivity from Prism Central to each PE VIP and consistent time sync; a cluster can be registered to only one PC.

Categories are the key mechanism for applying policies (protection, security, cost) consistently across clusters and clouds.

Cloud-native workloads can be managed alongside AHV VMs; Nutanix provides CSI/CNI integrations for Kubernetes (NKP) in multicloud scenarios.

Foundation discovers nodes via IPMI/Redfish out-of-band; it wipes disks, installs the hypervisor and AOS, and creates the cluster from a jump host.

Minimum 3 nodes for RF2, 4 nodes for RF3; single-node clusters exist but provide no redundancy.

Cluster creation requires cluster name, VIP, CVM IPs, hypervisor IPs, subnet mask, gateway, DNS, and NTP; the VIP must be a free IP in the CVM subnet.

Default CVM memory is 16 GB (32 GB for large nodes); the CVM reserves host memory, so size hosts for guest VMs plus CVM overhead.

AOS and hypervisor version compatibility is enforced by the compatibility matrix; mismatches block upgrades and cluster expansion.

After Foundation, run NCC health checks and confirm a healthy state before adding workloads; a new cluster starts with a default storage pool and container.

Prism Element manages one cluster; Prism Central is a separate VM (or scale-out set) that manages many clusters via registered Prism Element instances.

The cluster VIP, CVM IPs, and hypervisor management IPs must be unique and outside any DHCP pool used by guest VMs.

Prism Element shows cluster CPU, memory, storage, and IOPS for one cluster; Prism Central adds multi-cluster dashboards, categories, and policies.

NCC (Nutanix Cluster Check) runs a library of health checks; schedule it and triage failures - a failed check does not automatically mean an outage.

Alerts have severity levels (Critical, Warning, Info) and can route to email, SNMP, syslog, or webhooks; they can be acknowledged or resolved in Prism.

NTP drift is a common cause of cluster-wide issues; verify time sync across CVMs and hypervisors.

Pulse (formerly CallHome) sends telemetry to Nutanix support; disabling it limits proactive support and automated health insights.

LCM (Life Cycle Manager) handles AOS, hypervisor, firmware, and NCC updates with pre-checks and staged upgrades.

Cluster capacity is reported as usable vs. raw; RF2 consumes roughly 2x the data footprint plus overhead, so plan capacity with RF in mind.

A cluster can be registered to only one Prism Central at a time; PC must reach each PE VIP over the network.

The default bridge (br0) on each host carries CVM and hypervisor management traffic; guest VM traffic uses a separate bridge or VLAN-backed network.

AHV networks are VLAN-backed virtual networks; a VM NIC attaches to a network, and the VLAN ID defines the broadcast domain.

Bond modes: active-backup (default, no switch config), balance-slb (no LACP, source MAC hashing), and balance-tcp (requires LACP on the switch).

balance-tcp requires a correctly configured LACP port-channel upstream; misconfiguration causes connectivity loss and is a common deployment error.

The cluster VIP and CVM IPs belong on a management VLAN; isolate guest VM networks from management traffic using separate VLANs.

AHV uses Open vSwitch (OVS); Flow microsegmentation policies are defined in Prism Central using categories.

Diagnose with Prism network visualization, CVM ping/arp tests, and NCC network checks; failed CVM-to-CVM ping indicates a bridge or VLAN issue.

On ESXi clusters, the CVM uses a vSwitch port group; enable promiscuous mode and forged transmits on the dedicated port group.

A storage pool groups physical disks (HDD/SSD/NVMe); a container is a logical namespace (datastore) presented to the hypervisor, and multiple containers can share a pool.

Replication factor (RF) 2 tolerates one node/disk failure; RF3 tolerates two. RF is set per container or as a cluster default and consumes capacity accordingly.

Erasure coding (EC-X) reduces capacity overhead for cold data; it is configured per container and requires a minimum number of nodes (typically 4+).

The distributed storage fabric (DSF) is log-structured and extent-based; writes land in the local CVM staging area, then replicate to peers before acknowledgment.

Deduplication (capacity and/or inline) and compression are container-level settings; compression is inline by default on supported AOS versions.

Snapshots are metadata-based and space-efficient; they are not full copies, but changed blocks consume space over time.

Storage QoS (IOPS limits) can be applied per VM or per container to prevent noisy-neighbor effects.

Check storage health via the Prism storage dashboard, NCC disk checks, and ncli; a disk in a bad state triggers a rebuild that consumes cluster resources.

In AHV, VMs are created from images (ISOs, disk images) stored in a container; images can be uploaded via Prism, ncli, or the API.

Categories (key-value pairs) drive policy application in Prism Central for protection, recovery, and security; use categories to group VMs.

AHV supports live migration, HA (restart on host failure), and affinity/anti-affinity rules enforced by the scheduler.

VM affinity rules keep VMs on the same host; anti-affinity rules spread them across hosts. Rules can be 'should' or 'must' strength.

Prism Central Projects provide self-service VM provisioning with quotas, role-based access, and environment isolation.

Nutanix Guest Tools (NGT) enables file-level restore, application-consistent snapshots, and self-service restore for supported guest OSes.

VM power operations, console access, and snapshot/clone actions are available in Prism Element; bulk operations and policies are in Prism Central.

A VM clone can be full or linked (dependent on the parent snapshot); linked clones save space but depend on the parent image/snapshot chain.

Protection domains group VMs (by category) for replication and snapshot policies; a protection policy defines RPO, retention, and replication target.

Async replication sends snapshots to a remote cluster at the RPO interval; sync replication (Metro Availability) requires low-latency links and provides near-zero RPO.

Nutanix DR uses protection policies in Prism Central; a recovery plan defines boot order and network mapping for failover.

Snapshots are local and fast; replication is remote. A protection policy can take local snapshots only, or local plus remote replication.

NearSync replication provides RPO in seconds (typically 1-5 minutes) and requires a compatible AOS version and network latency budget.

For file-level restore, NGT must be installed in the guest; self-service restore lets users recover their own files from a snapshot.

Test failover (planned) is non-disruptive; unplanned failover promotes the remote copy. Validate the recovery plan before relying on it.

Retention policies define how many snapshots/replicas to keep (e.g., hourly/daily/weekly); retention applies per protection policy and consumes capacity on both sites.

NCC is the first diagnostic step: run it from the CVM CLI or via Prism; failed checks often point to the root cause (time sync, disk, network).

CVM memory pressure is a common cause of cluster instability; monitor CVM memory and resize if consistently high.

A disk failure triggers a rebuild; monitor rebuild progress in Prism and avoid maintenance operations during a rebuild.

NTP drift causes authentication, replication, and cluster services failures; verify NTP on CVMs and hypervisors with ntpq -p or chronyc sources.

Log collection: generate a log bundle via ncli or Prism; logs are also on the CVM under /home/nutanix/data/logs.

LCM pre-checks must pass before upgrades; common blockers are insufficient disk space, NCC failures, and version incompatibility.

A node in maintenance mode evacuates VMs and data; use it before hardware service to avoid data loss and keep RF satisfied.

Cluster health is not just green lights - check the health dashboard, NCC, and capacity trends; a cluster can be functional but degraded (e.g., RF not satisfied).

Prism Central is the multicloud control plane; it manages multiple Prism Element clusters (on-prem and in public cloud) from one UI.

Nutanix Clusters on AWS/Azure extends AOS networking and management to public cloud; it is not the same as running AHV on-prem.

Calm (Nutanix Cloud Manager) provides orchestration and self-service for multi-cloud workloads; blueprints define multi-tier applications.

Nutanix Cloud Manager (NCM) includes cost governance, security, and self-service; it integrates with Prism Central and public cloud accounts.

Flow (microsegmentation) policies are defined in Prism Central and applied across clusters; they use categories, not IP addresses, for policy targets.

Multi-cluster management requires network connectivity from Prism Central to each PE VIP and consistent time sync; a cluster can be registered to only one PC.

Categories are the key mechanism for applying policies (protection, security, cost) consistently across clusters and clouds.

Cloud-native workloads can be managed alongside AHV VMs; Nutanix provides CSI/CNI integrations for Kubernetes (NKP) in multicloud scenarios.

Foundation discovers nodes via IPMI/Redfish out-of-band; it wipes disks, installs the hypervisor and AOS, and creates the cluster from a jump host.

Minimum 3 nodes for RF2, 4 nodes for RF3; single-node clusters exist but provide no redundancy.

Cluster creation requires cluster name, VIP, CVM IPs, hypervisor IPs, subnet mask, gateway, DNS, and NTP; the VIP must be a free IP in the CVM subnet.

Default CVM memory is 16 GB (32 GB for large nodes); the CVM reserves host memory, so size hosts for guest VMs plus CVM overhead.

AOS and hypervisor version compatibility is enforced by the compatibility matrix; mismatches block upgrades and cluster expansion.

After Foundation, run NCC health checks and confirm a healthy state before adding workloads; a new cluster starts with a default storage pool and container.

Prism Element manages one cluster; Prism Central is a separate VM (or scale-out set) that manages many clusters via registered Prism Element instances.

The cluster VIP, CVM IPs, and hypervisor management IPs must be unique and outside any DHCP pool used by guest VMs.

Prism Element shows cluster CPU, memory, storage, and IOPS for one cluster; Prism Central adds multi-cluster dashboards, categories, and policies.

NCC (Nutanix Cluster Check) runs a library of health checks; schedule it and triage failures - a failed check does not automatically mean an outage.

Alerts have severity levels (Critical, Warning, Info) and can route to email, SNMP, syslog, or webhooks; they can be acknowledged or resolved in Prism.

NTP drift is a common cause of cluster-wide issues; verify time sync across CVMs and hypervisors.

Pulse (formerly CallHome) sends telemetry to Nutanix support; disabling it limits proactive support and automated health insights.

LCM (Life Cycle Manager) handles AOS, hypervisor, firmware, and NCC updates with pre-checks and staged upgrades.

Cluster capacity is reported as usable vs. raw; RF2 consumes roughly 2x the data footprint plus overhead, so plan capacity with RF in mind.

A cluster can be registered to only one Prism Central at a time; PC must reach each PE VIP over the network.

The default bridge (br0) on each host carries CVM and hypervisor management traffic; guest VM traffic uses a separate bridge or VLAN-backed network.

AHV networks are VLAN-backed virtual networks; a VM NIC attaches to a network, and the VLAN ID defines the broadcast domain.

Bond modes: active-backup (default, no switch config), balance-slb (no LACP, source MAC hashing), and balance-tcp (requires LACP on the switch).

balance-tcp requires a correctly configured LACP port-channel upstream; misconfiguration causes connectivity loss and is a common deployment error.

The cluster VIP and CVM IPs belong on a management VLAN; isolate guest VM networks from management traffic using separate VLANs.

AHV uses Open vSwitch (OVS); Flow microsegmentation policies are defined in Prism Central using categories.

Diagnose with Prism network visualization, CVM ping/arp tests, and NCC network checks; failed CVM-to-CVM ping indicates a bridge or VLAN issue.

On ESXi clusters, the CVM uses a vSwitch port group; enable promiscuous mode and forged transmits on the dedicated port group.

A storage pool groups physical disks (HDD/SSD/NVMe); a container is a logical namespace (datastore) presented to the hypervisor, and multiple containers can share a pool.

Replication factor (RF) 2 tolerates one node/disk failure; RF3 tolerates two. RF is set per container or as a cluster default and consumes capacity accordingly.

Erasure coding (EC-X) reduces capacity overhead for cold data; it is configured per container and requires a minimum number of nodes (typically 4+).

The distributed storage fabric (DSF) is log-structured and extent-based; writes land in the local CVM staging area, then replicate to peers before acknowledgment.

Deduplication (capacity and/or inline) and compression are container-level settings; compression is inline by default on supported AOS versions.

Snapshots are metadata-based and space-efficient; they are not full copies, but changed blocks consume space over time.

Storage QoS (IOPS limits) can be applied per VM or per container to prevent noisy-neighbor effects.

Check storage health via the Prism storage dashboard, NCC disk checks, and ncli; a disk in a bad state triggers a rebuild that consumes cluster resources.

In AHV, VMs are created from images (ISOs, disk images) stored in a container; images can be uploaded via Prism, ncli, or the API.

Categories (key-value pairs) drive policy application in Prism Central for protection, recovery, and security; use categories to group VMs.

AHV supports live migration, HA (restart on host failure), and affinity/anti-affinity rules enforced by the scheduler.

VM affinity rules keep VMs on the same host; anti-affinity rules spread them across hosts. Rules can be 'should' or 'must' strength.

Prism Central Projects provide self-service VM provisioning with quotas, role-based access, and environment isolation.

Nutanix Guest Tools (NGT) enables file-level restore, application-consistent snapshots, and self-service restore for supported guest OSes.

VM power operations, console access, and snapshot/clone actions are available in Prism Element; bulk operations and policies are in Prism Central.

A VM clone can be full or linked (dependent on the parent snapshot); linked clones save space but depend on the parent image/snapshot chain.

Protection domains group VMs (by category) for replication and snapshot policies; a protection policy defines RPO, retention, and replication target.

Async replication sends snapshots to a remote cluster at the RPO interval; sync replication (Metro Availability) requires low-latency links and provides near-zero RPO.

Nutanix DR uses protection policies in Prism Central; a recovery plan defines boot order and network mapping for failover.

Snapshots are local and fast; replication is remote. A protection policy can take local snapshots only, or local plus remote replication.

NearSync replication provides RPO in seconds (typically 1-5 minutes) and requires a compatible AOS version and network latency budget.

For file-level restore, NGT must be installed in the guest; self-service restore lets users recover their own files from a snapshot.

Test failover (planned) is non-disruptive; unplanned failover promotes the remote copy. Validate the recovery plan before relying on it.

Retention policies define how many snapshots/replicas to keep (e.g., hourly/daily/weekly); retention applies per protection policy and consumes capacity on both sites.

NCC is the first diagnostic step: run it from the CVM CLI or via Prism; failed checks often point to the root cause (time sync, disk, network).

CVM memory pressure is a common cause of cluster instability; monitor CVM memory and resize if consistently high.

A disk failure triggers a rebuild; monitor rebuild progress in Prism and avoid maintenance operations during a rebuild.

NTP drift causes authentication, replication, and cluster services failures; verify NTP on CVMs and hypervisors with ntpq -p or chronyc sources.

Log collection: generate a log bundle via ncli or Prism; logs are also on the CVM under /home/nutanix/data/logs.

LCM pre-checks must pass before upgrades; common blockers are insufficient disk space, NCC failures, and version incompatibility.

A node in maintenance mode evacuates VMs and data; use it before hardware service to avoid data loss and keep RF satisfied.

Cluster health is not just green lights - check the health dashboard, NCC, and capacity trends; a cluster can be functional but degraded (e.g., RF not satisfied).

Prism Central is the multicloud control plane; it manages multiple Prism Element clusters (on-prem and in public cloud) from one UI.

Nutanix Clusters on AWS/Azure extends AOS networking and management to public cloud; it is not the same as running AHV on-prem.

Calm (Nutanix Cloud Manager) provides orchestration and self-service for multi-cloud workloads; blueprints define multi-tier applications.

Nutanix Cloud Manager (NCM) includes cost governance, security, and self-service; it integrates with Prism Central and public cloud accounts.

Flow (microsegmentation) policies are defined in Prism Central and applied across clusters; they use categories, not IP addresses, for policy targets.

Multi-cluster management requires network connectivity from Prism Central to each PE VIP and consistent time sync; a cluster can be registered to only one PC.

Categories are the key mechanism for applying policies (protection, security, cost) consistently across clusters and clouds.

Cloud-native workloads can be managed alongside AHV VMs; Nutanix provides CSI/CNI integrations for Kubernetes (NKP) in multicloud scenarios.

Foundation discovers nodes via IPMI/Redfish out-of-band; it wipes disks, installs the hypervisor and AOS, and creates the cluster from a jump host.

Minimum 3 nodes for RF2, 4 nodes for RF3; single-node clusters exist but provide no redundancy.

Cluster creation requires cluster name, VIP, CVM IPs, hypervisor IPs, subnet mask, gateway, DNS, and NTP; the VIP must be a free IP in the CVM subnet.

Default CVM memory is 16 GB (32 GB for large nodes); the CVM reserves host memory, so size hosts for guest VMs plus CVM overhead.

AOS and hypervisor version compatibility is enforced by the compatibility matrix; mismatches block upgrades and cluster expansion.

After Foundation, run NCC health checks and confirm a healthy state before adding workloads; a new cluster starts with a default storage pool and container.

Prism Element manages one cluster; Prism Central is a separate VM (or scale-out set) that manages many clusters via registered Prism Element instances.

The cluster VIP, CVM IPs, and hypervisor management IPs must be unique and outside any DHCP pool used by guest VMs.

Prism Element shows cluster CPU, memory, storage, and IOPS for one cluster; Prism Central adds multi-cluster dashboards, categories, and policies.

NCC (Nutanix Cluster Check) runs a library of health checks; schedule it and triage failures - a failed check does not automatically mean an outage.

Alerts have severity levels (Critical, Warning, Info) and can route to email, SNMP, syslog, or webhooks; they can be acknowledged or resolved in Prism.

NTP drift is a common cause of cluster-wide issues; verify time sync across CVMs and hypervisors.

Pulse (formerly CallHome) sends telemetry to Nutanix support; disabling it limits proactive support and automated health insights.

LCM (Life Cycle Manager) handles AOS, hypervisor, firmware, and NCC updates with pre-checks and staged upgrades.

Cluster capacity is reported as usable vs. raw; RF2 consumes roughly 2x the data footprint plus overhead, so plan capacity with RF in mind.

A cluster can be registered to only one Prism Central at a time; PC must reach each PE VIP over the network.

The default bridge (br0) on each host carries CVM and hypervisor management traffic; guest VM traffic uses a separate bridge or VLAN-backed network.

AHV networks are VLAN-backed virtual networks; a VM NIC attaches to a network, and the VLAN ID defines the broadcast domain.

Bond modes: active-backup (default, no switch config), balance-slb (no LACP, source MAC hashing), and balance-tcp (requires LACP on the switch).

balance-tcp requires a correctly configured LACP port-channel upstream; misconfiguration causes connectivity loss and is a common deployment error.

The cluster VIP and CVM IPs belong on a management VLAN; isolate guest VM networks from management traffic using separate VLANs.

AHV uses Open vSwitch (OVS); Flow microsegmentation policies are defined in Prism Central using categories.

Diagnose with Prism network visualization, CVM ping/arp tests, and NCC network checks; failed CVM-to-CVM ping indicates a bridge or VLAN issue.

On ESXi clusters, the CVM uses a vSwitch port group; enable promiscuous mode and forged transmits on the dedicated port group.

A storage pool groups physical disks (HDD/SSD/NVMe); a container is a logical namespace (datastore) presented to the hypervisor, and multiple containers can share a pool.

Replication factor (RF) 2 tolerates one node/disk failure; RF3 tolerates two. RF is set per container or as a cluster default and consumes capacity accordingly.

Erasure coding (EC-X) reduces capacity overhead for cold data; it is configured per container and requires a minimum number of nodes (typically 4+).

The distributed storage fabric (DSF) is log-structured and extent-based; writes land in the local CVM staging area, then replicate to peers before acknowledgment.

Deduplication (capacity and/or inline) and compression are container-level settings; compression is inline by default on supported AOS versions.

Snapshots are metadata-based and space-efficient; they are not full copies, but changed blocks consume space over time.

Storage QoS (IOPS limits) can be applied per VM or per container to prevent noisy-neighbor effects.

Check storage health via the Prism storage dashboard, NCC disk checks, and ncli; a disk in a bad state triggers a rebuild that consumes cluster resources.

In AHV, VMs are created from images (ISOs, disk images) stored in a container; images can be uploaded via Prism, ncli, or the API.

Categories (key-value pairs) drive policy application in Prism Central for protection, recovery, and security; use categories to group VMs.

AHV supports live migration, HA (restart on host failure), and affinity/anti-affinity rules enforced by the scheduler.

VM affinity rules keep VMs on the same host; anti-affinity rules spread them across hosts. Rules can be 'should' or 'must' strength.

Prism Central Projects provide self-service VM provisioning with quotas, role-based access, and environment isolation.

Nutanix Guest Tools (NGT) enables file-level restore, application-consistent snapshots, and self-service restore for supported guest OSes.

VM power operations, console access, and snapshot/clone actions are available in Prism Element; bulk operations and policies are in Prism Central.

A VM clone can be full or linked (dependent on the parent snapshot); linked clones save space but depend on the parent image/snapshot chain.

Protection domains group VMs (by category) for replication and snapshot policies; a protection policy defines RPO, retention, and replication target.

Async replication sends snapshots to a remote cluster at the RPO interval; sync replication (Metro Availability) requires low-latency links and provides near-zero RPO.

Nutanix DR uses protection policies in Prism Central; a recovery plan defines boot order and network mapping for failover.

Snapshots are local and fast; replication is remote. A protection policy can take local snapshots only, or local plus remote replication.

NearSync replication provides RPO in seconds (typically 1-5 minutes) and requires a compatible AOS version and network latency budget.

For file-level restore, NGT must be installed in the guest; self-service restore lets users recover their own files from a snapshot.

Test failover (planned) is non-disruptive; unplanned failover promotes the remote copy. Validate the recovery plan before relying on it.

Retention policies define how many snapshots/replicas to keep (e.g., hourly/daily/weekly); retention applies per protection policy and consumes capacity on both sites.

NCC is the first diagnostic step: run it from the CVM CLI or via Prism; failed checks often point to the root cause (time sync, disk, network).

CVM memory pressure is a common cause of cluster instability; monitor CVM memory and resize if consistently high.

A disk failure triggers a rebuild; monitor rebuild progress in Prism and avoid maintenance operations during a rebuild.

NTP drift causes authentication, replication, and cluster services failures; verify NTP on CVMs and hypervisors with ntpq -p or chronyc sources.

Log collection: generate a log bundle via ncli or Prism; logs are also on the CVM under /home/nutanix/data/logs.

LCM pre-checks must pass before upgrades; common blockers are insufficient disk space, NCC failures, and version incompatibility.

A node in maintenance mode evacuates VMs and data; use it before hardware service to avoid data loss and keep RF satisfied.

Cluster health is not just green lights - check the health dashboard, NCC, and capacity trends; a cluster can be functional but degraded (e.g., RF not satisfied).

Prism Central is the multicloud control plane; it manages multiple Prism Element clusters (on-prem and in public cloud) from one UI.

Nutanix Clusters on AWS/Azure extends AOS networking and management to public cloud; it is not the same as running AHV on-prem.

Calm (Nutanix Cloud Manager) provides orchestration and self-service for multi-cloud workloads; blueprints define multi-tier applications.

Nutanix Cloud Manager (NCM) includes cost governance, security, and self-service; it integrates with Prism Central and public cloud accounts.

Flow (microsegmentation) policies are defined in Prism Central and applied across clusters; they use categories, not IP addresses, for policy targets.

Multi-cluster management requires network connectivity from Prism Central to each PE VIP and consistent time sync; a cluster can be registered to only one PC.

Categories are the key mechanism for applying policies (protection, security, cost) consistently across clusters and clouds.

Cloud-native workloads can be managed alongside AHV VMs; Nutanix provides CSI/CNI integrations for Kubernetes (NKP) in multicloud scenarios.

Foundation discovers nodes via IPMI/Redfish out-of-band; it wipes disks, installs the hypervisor and AOS, and creates the cluster from a jump host.

Minimum 3 nodes for RF2, 4 nodes for RF3; single-node clusters exist but provide no redundancy.

Cluster creation requires cluster name, VIP, CVM IPs, hypervisor IPs, subnet mask, gateway, DNS, and NTP; the VIP must be a free IP in the CVM subnet.

Default CVM memory is 16 GB (32 GB for large nodes); the CVM reserves host memory, so size hosts for guest VMs plus CVM overhead.

AOS and hypervisor version compatibility is enforced by the compatibility matrix; mismatches block upgrades and cluster expansion.

After Foundation, run NCC health checks and confirm a healthy state before adding workloads; a new cluster starts with a default storage pool and container.

Prism Element manages one cluster; Prism Central is a separate VM (or scale-out set) that manages many clusters via registered Prism Element instances.

The cluster VIP, CVM IPs, and hypervisor management IPs must be unique and outside any DHCP pool used by guest VMs.

Prism Element shows cluster CPU, memory, storage, and IOPS for one cluster; Prism Central adds multi-cluster dashboards, categories, and policies.

NCC (Nutanix Cluster Check) runs a library of health checks; schedule it and triage failures - a failed check does not automatically mean an outage.

Alerts have severity levels (Critical, Warning, Info) and can route to email, SNMP, syslog, or webhooks; they can be acknowledged or resolved in Prism.

NTP drift is a common cause of cluster-wide issues; verify time sync across CVMs and hypervisors.

Pulse (formerly CallHome) sends telemetry to Nutanix support; disabling it limits proactive support and automated health insights.

LCM (Life Cycle Manager) handles AOS, hypervisor, firmware, and NCC updates with pre-checks and staged upgrades.

Cluster capacity is reported as usable vs. raw; RF2 consumes roughly 2x the data footprint plus overhead, so plan capacity with RF in mind.

A cluster can be registered to only one Prism Central at a time; PC must reach each PE VIP over the network.

The default bridge (br0) on each host carries CVM and hypervisor management traffic; guest VM traffic uses a separate bridge or VLAN-backed network.

AHV networks are VLAN-backed virtual networks; a VM NIC attaches to a network, and the VLAN ID defines the broadcast domain.

Bond modes: active-backup (default, no switch config), balance-slb (no LACP, source MAC hashing), and balance-tcp (requires LACP on the switch).

balance-tcp requires a correctly configured LACP port-channel upstream; misconfiguration causes connectivity loss and is a common deployment error.

The cluster VIP and CVM IPs belong on a management VLAN; isolate guest VM networks from management traffic using separate VLANs.

AHV uses Open vSwitch (OVS); Flow microsegmentation policies are defined in Prism Central using categories.

Diagnose with Prism network visualization, CVM ping/arp tests, and NCC network checks; failed CVM-to-CVM ping indicates a bridge or VLAN issue.

On ESXi clusters, the CVM uses a vSwitch port group; enable promiscuous mode and forged transmits on the dedicated port group.

A storage pool groups physical disks (HDD/SSD/NVMe); a container is a logical namespace (datastore) presented to the hypervisor, and multiple containers can share a pool.

Replication factor (RF) 2 tolerates one node/disk failure; RF3 tolerates two. RF is set per container or as a cluster default and consumes capacity accordingly.

Erasure coding (EC-X) reduces capacity overhead for cold data; it is configured per container and requires a minimum number of nodes (typically 4+).

The distributed storage fabric (DSF) is log-structured and extent-based; writes land in the local CVM staging area, then replicate to peers before acknowledgment.

Deduplication (capacity and/or inline) and compression are container-level settings; compression is inline by default on supported AOS versions.

Snapshots are metadata-based and space-efficient; they are not full copies, but changed blocks consume space over time.

Storage QoS (IOPS limits) can be applied per VM or per container to prevent noisy-neighbor effects.

Check storage health via the Prism storage dashboard, NCC disk checks, and ncli; a disk in a bad state triggers a rebuild that consumes cluster resources.

In AHV, VMs are created from images (ISOs, disk images) stored in a container; images can be uploaded via Prism, ncli, or the API.

Categories (key-value pairs) drive policy application in Prism Central for protection, recovery, and security; use categories to group VMs.

AHV supports live migration, HA (restart on host failure), and affinity/anti-affinity rules enforced by the scheduler.

VM affinity rules keep VMs on the same host; anti-affinity rules spread them across hosts. Rules can be 'should' or 'must' strength.

Prism Central Projects provide self-service VM provisioning with quotas, role-based access, and environment isolation.

Nutanix Guest Tools (NGT) enables file-level restore, application-consistent snapshots, and self-service restore for supported guest OSes.

VM power operations, console access, and snapshot/clone actions are available in Prism Element; bulk operations and policies are in Prism Central.

A VM clone can be full or linked (dependent on the parent snapshot); linked clones save space but depend on the parent image/snapshot chain.

Protection domains group VMs (by category) for replication and snapshot policies; a protection policy defines RPO, retention, and replication target.

Async replication sends snapshots to a remote cluster at the RPO interval; sync replication (Metro Availability) requires low-latency links and provides near-zero RPO.

Nutanix DR uses protection policies in Prism Central; a recovery plan defines boot order and network mapping for failover.

Snapshots are local and fast; replication is remote. A protection policy can take local snapshots only, or local plus remote replication.

NearSync replication provides RPO in seconds (typically 1-5 minutes) and requires a compatible AOS version and network latency budget.

For file-level restore, NGT must be installed in the guest; self-service restore lets users recover their own files from a snapshot.

Test failover (planned) is non-disruptive; unplanned failover promotes the remote copy. Validate the recovery plan before relying on it.

Retention policies define how many snapshots/replicas to keep (e.g., hourly/daily/weekly); retention applies per protection policy and consumes capacity on both sites.

NCC is the first diagnostic step: run it from the CVM CLI or via Prism; failed checks often point to the root cause (time sync, disk, network).

CVM memory pressure is a common cause of cluster instability; monitor CVM memory and resize if consistently high.

A disk failure triggers a rebuild; monitor rebuild progress in Prism and avoid maintenance operations during a rebuild.

NTP drift causes authentication, replication, and cluster services failures; verify NTP on CVMs and hypervisors with ntpq -p or chronyc sources.

Log collection: generate a log bundle via ncli or Prism; logs are also on the CVM under /home/nutanix/data/logs.

LCM pre-checks must pass before upgrades; common blockers are insufficient disk space, NCC failures, and version incompatibility.

A node in maintenance mode evacuates VMs and data; use it before hardware service to avoid data loss and keep RF satisfied.

Cluster health is not just green lights - check the health dashboard, NCC, and capacity trends; a cluster can be functional but degraded (e.g., RF not satisfied).

Prism Central is the multicloud control plane; it manages multiple Prism Element clusters (on-prem and in public cloud) from one UI.

Nutanix Clusters on AWS/Azure extends AOS networking and management to public cloud; it is not the same as running AHV on-prem.

Calm (Nutanix Cloud Manager) provides orchestration and self-service for multi-cloud workloads; blueprints define multi-tier applications.

Nutanix Cloud Manager (NCM) includes cost governance, security, and self-service; it integrates with Prism Central and public cloud accounts.

Flow (microsegmentation) policies are defined in Prism Central and applied across clusters; they use categories, not IP addresses, for policy targets.

Multi-cluster management requires network connectivity from Prism Central to each PE VIP and consistent time sync; a cluster can be registered to only one PC.

Categories are the key mechanism for applying policies (protection, security, cost) consistently across clusters and clouds.

Cloud-native workloads can be managed alongside AHV VMs; Nutanix provides CSI/CNI integrations for Kubernetes (NKP) in multicloud scenarios.