This is on the exam!
NCA is foundational: terminology, what-does-what, and the resiliency numbers. The densest, most-tested ground is Section 4 (storage, RF/FT) and the LCM workflow.
1 · Describe Lifecycle Management LCM
Which tool updates both software and firmware from one workflow?
Right answerLCM (Life Cycle Manager). It is the unified, one-click updater for cluster software and firmware, and it resolves dependencies and sequencing for you.
The classic trap is a distractor that only does firmware or only does software. LCM does both.
What must you run before LCM can apply updates?
Right answerAn Inventory operation. It scans the cluster for installed component versions and populates the available updates.
Order: Inventory → review → Update. LCM then plans a safe rolling sequence (nodes may reboot).
How do you run LCM on a cluster with no internet (dark site)?
Right answerHost the LCM dark-site bundle on a local web server inside your network and point LCM at that URL.
The capability is not lost without Support Portal connectivity; only the source of the bundle changes.
Name things LCM can update.
Right answerNutanix software (AOS, Foundation, Prism, NCC/Licensing) and firmware (BIOS/BMC, disks, HBA, NIC) across Nutanix NX and qualified OEMs (Dell, HPE, Cisco, Lenovo, Fujitsu) and NC2.
Prism Element rides along with AOS; Prism Central is a separate component with its own upgrade path.
Why does LCM resolve dependencies itself?
Right answerSo you don't hand-pick firmware/software order. LCM computes a compatible upgrade sequence in advance, reducing manual risk.
This is the value proposition vs manually upgrading each component; it works even at dark sites.
2 · Describe Nutanix Basic Administration VM ops · Prism · licensing
Is AHV a separately licensed hypervisor?
Right answerNo. AHV is included at no extra licensing cost in every NCI license, and a new node ships with AHV by default.
AHV is KVM-based, hardened and operationalized by Nutanix. VM ops, live migration, HA, and virtual networking are built in.
Host fails unexpectedly. Live migration or HA, and is it a move or a restart?
Right answerHA, and it is a restart. VMs that were on the failed host are restarted on surviving hosts.
Trap vs Live Migration, which is a planned, no-downtime move of a running VM (memory copied while it runs).
Which feature auto-balances VM load and avoids hot spots?
Right answerADS (Acropolis Dynamic Scheduling). Automatic, on by default on every AHV cluster.
The AHV analog of VMware DRS. ADS-triggered moves use live migration under the hood.
Manage many clusters across sites from one screen, PE or PC?
Right answerPrism Central (PC), a separate VM (or scale-out VM cluster) you deploy to manage multiple clusters.
Prism Element (PE) is built into AOS, runs in every CVM, and manages only its own cluster. PE upgrades with AOS; PC is separate.
Does Prism Element cost extra or need a separate deployment?
Right answerNo. PE is built into every cluster (runs inside the CVMs). Prism Central is the piece you deploy separately.
Prism Starter licensing is included with every AOS license; a 90-day Prism Ultimate trial is included too.
Use case needs Metro / NearSync / encryption / microsegmentation, which AOS tier?
Right answerUltimate. Advanced replication (Metro, NearSync, cross-cluster migration), encryption (data-at-rest, native KMS), and Flow microsegmentation are Ultimate features.
RF3 / erasure coding / >12 nodes → Pro. Basic RF2 + async + compression → Starter.
Which AOS tier caps at 12 nodes and offers only RF2?
Right answerStarter (up to 12 nodes per cluster, RF2, compression/dedup, async ≥ 1 hour, LCM).
RF3 and erasure coding require Pro, which also lifts the node cap above 12.
CLI: which utility for VM/network ops vs cluster ops?
Right answeracli (Acropolis CLI) for VM and network operations; ncli (Nutanix CLI) for cluster-level operations.
Most day-to-day work is in Prism; the CLIs appear in "which command/where" style questions.
3 · Maintain Environmental Health NCC · Pulse · alerts
Difference between a health check and an alert?
Right answerChecks are scheduled/proactive (run on a schedule via NCC); alerts are event-driven/reactive (raised when something crosses a threshold or fails).
Alerts carry severities (info / warning / critical) and can notify out by email.
What is NCC?
Right answerNutanix Cluster Check, the health-check framework that scans configuration, performance, and operational issues across the cluster. Run from Prism or ncc health_checks run_all.
NCC is downloaded/updated from the Support Portal, and Pulse is packaged with it.
Which feature sends cluster telemetry to Nutanix for proactive support?
Right answerPulse. When enabled it securely sends cluster health and configuration data to Nutanix on a schedule.
Pulse feeds Nutanix Insights for predictive health and faster support. Pulse = scheduled telemetry; alerts = event-driven.
Where do you see overall cluster status and data resiliency at a glance?
Right answerThe Health dashboard in Prism (hosts, disks, VMs, storage, and the resiliency state).
The data-resiliency widget tells you whether the cluster can currently tolerate its configured failures.
What is the Support Portal used for?
Right answerDownload software (AOS, NCC, Foundation, LCM dark-site bundles), manage licenses, open/track support cases, and read docs/KB.
Pulse data surfaces in the portal's Insights views to speed triage.
4 · Describe Cluster Configuration Options storage · RF/FT · network
Minimum nodes for RF2? For RF3?
Right answerRF2 needs 3 nodes; RF3 needs 5 nodes.
RF3 keeps 5 copies of metadata, which is why it needs 5 nodes even though data has 3 copies.
How many failures does each redundancy level tolerate?
Right answerRF2 / FT1 tolerates 1 node-or-drive failure; RF3 / FT2 tolerates 2 simultaneous failures.
FT1 ↔ data RF2 (metadata RF3); FT2 ↔ data RF3 (metadata RF5). Cluster default is RF2 / FT1.
Where are RF and storage optimization set?
Right answerOn the storage container (RF, compression, deduplication, erasure coding are all container-level policies).
The storage pool is the raw capacity of all disks; the container is the logical slice where policy lives; vDisks live in containers.
Which optimization fits VDI / full clones?
Right answerDeduplication (SHA-1 fingerprint on ingest; removes duplicate blocks). Ideal where many VMs share identical data.
Dedup works inline and post-process across the performance and capacity tiers.
Which optimization fits cold, rarely-changed data?
Right answerErasure Coding (EC-X). RAID-like parity striped across nodes; post-process on data cold for >7 days.
It reclaims capacity without RF's full-copy overhead, so it suits write-cold/archival data, not hot, frequently overwritten data.
Default / safest general storage optimization?
Right answerCompression. Inline for large/sequential I/O (>64 KB) or post-process; LZ4 / LZ4HC.
Broadly safe to enable across workloads; the usual default recommendation.
Which hypervisors does Nutanix support?
Right answerAHV (built in, free), VMware ESXi, and Microsoft Hyper-V. AHV is the default.
Foundation images the chosen hypervisor at cluster build time.
What is NC2 and where does it run?
Right answerNutanix Cloud Clusters, the full Nutanix stack (CVM + AHV + AOS + Prism) on bare-metal instances in AWS and Azure.
It is the "extend or migrate to public cloud without changing the operating model" answer; operated as one cloud with on-prem.
What does a two-node (ROBO) cluster require that a three-node does not?
Right answerAn external Witness VM in a separate failure domain (to arbitrate and keep resiliency).
3 nodes is the standard minimum. 1-node ROBO exists on selected hardware with reduced resiliency and a limited guest-VM count.
What serves storage on every node?
Right answerThe Controller VM (CVM), which runs the Distributed Storage Fabric (DSF) and presents pooled storage to the hypervisor.
Data locality keeps a VM's hot data on the node where it runs.
AHV networking: virtual switch and default bridge?
Right answerOpen vSwitch (OVS), default bridge br0 with default uplink bond br0-up.
Put the CVM and AHV host on the native/untagged VLAN; a bond should have at least two physical NICs for HA. OVS supports VLAN tagging and LACP.
Which tool images bare-metal nodes and forms a cluster?
Right answerFoundation. It installs the chosen AOS + hypervisor and forms the cluster; built into the CVMs.
Plan ~3 IPs per node (hypervisor, CVM, IPMI/remote mgmt) plus a cluster IP and a Data Services IP.