I85PREP
⌘ K
NCP-ADS study materials
NCP-ADS study resources

NCP-ADS practice questions

One question at a time. Build your NCP-ADS understanding.

First 50 of 200 questions. Each one lists the choices, the correct answer, and an explanation. The checker at the top of this page saves progress in this browser when it is available.

Continue in the practice tool above to load more questions, 50 at a time.

Question 1Data Center and Networking Fundamentals

In a data center, which type of server is designed primarily to host many independent virtual machines or containers on a single physical chassis?

  1. A blade server with shared backplane power and cooling
  2. A network-attached storage head unit
  3. A dedicated bare-metal database appliance
  4. A virtualization host with high core count and large memory capacity Correct answer.

Explanation

A virtualization host is provisioned with high CPU core counts, large memory capacity, and multiple network interfaces so that many guest workloads can share the hardware. Blade servers describe a physical form factor rather than the virtualization role, and storage heads and database appliances serve narrower purposes.

Question 2Data Center and Networking Fundamentals

Which cabling medium is the standard choice for a 10GBASE-T link within a data center rack when distance stays under 100 meters?

  1. Category 6A twisted pair Correct answer.
  2. Category 5e unshielded twisted pair
  3. Multi-mode fiber with LC connectors
  4. Twinaxial direct-attach copper

Explanation

10GBASE-T over copper requires Category 6A (or better) to support the full 100-meter channel, because Cat 6A is specified to higher frequencies with reduced alien crosstalk. Category 5e is limited to 1 Gbps at that distance, while fiber and twinax use different physical interfaces.

Question 3Data Center and Networking Fundamentals

What is the primary purpose of a top-of-rack (ToR) switch in a data center design?

  1. To aggregate server connections within a single rack and uplink to the aggregation layer Correct answer.
  2. To provide out-of-band console access to servers
  3. To distribute electrical power to rack-mounted equipment
  4. To terminate wide-area circuits from the service provider

Explanation

A ToR switch sits at the top of a rack, connects the servers in that rack at access-layer speeds, and uplinks northbound to aggregation or spine devices. This shortens cable runs and simplifies rack-level cabling.

Question 4Data Center and Networking Fundamentals

Which address range is reserved for IPv4 link-local autoconfiguration (APIPA) and is therefore unsuitable for routable data center addressing?

  1. 192.168.0.0/16
  2. 172.16.0.0/12
  3. 169.254.0.0/16 Correct answer.
  4. 10.0.0.0/8

Explanation

The 169.254.0.0/16 block is assigned for link-local addressing, used when DHCP is unavailable, and routers do not forward it. The other three ranges are private RFC 1918 space that is commonly routed inside enterprise and data center networks.

Question 5Data Center and Networking Fundamentals

A server must keep operating through the failure of one power supply while still receiving power from the utility feed. Which configuration meets this requirement?

  1. Dual PSUs with one connected to feed A and the other to feed B Correct answer.
  2. A single PSU connected to a single rack PDU on feed A
  3. A single PSU connected to a UPS output only
  4. Dual PSUs both connected to the same PDU on feed A

Explanation

Redundant power requires each PSU to draw from an independent power path, typically A and B feeds backed by separate UPS systems. If both PSUs share one PDU or feed, a single failure at that point still takes the server down.

Question 6Data Center and Networking Fundamentals

Which protocol provides a loop-free Layer 2 topology by electing a root bridge and blocking redundant ports?

  1. LACP
  2. BGP
  3. Spanning Tree Protocol Correct answer.
  4. OSPF

Explanation

Spanning Tree Protocol builds a loop-free logical tree by electing a root bridge and placing redundant ports into a blocking state. OSPF and BGP are Layer 3 routing protocols, and LACP bundles links rather than preventing Layer 2 loops.

Question 7Data Center and Networking Fundamentals

What does a /26 IPv4 subnet mask provide in terms of usable host addresses?

  1. 254 usable hosts
  2. 30 usable hosts
  3. 62 usable hosts Correct answer.
  4. 126 usable hosts

Explanation

A /26 leaves 6 host bits, giving 64 addresses; subtracting the network and broadcast addresses yields 62 usable hosts. The other counts correspond to /27, /25, and /24 respectively.

Question 8Data Center and Networking Fundamentals

Which device operates at Layer 3 of the OSI model and makes forwarding decisions based on destination IP address?

  1. Layer 2 access switch
  2. Router Correct answer.
  3. Patch panel
  4. Media converter

Explanation

A router examines the destination IP address and consults a routing table to forward packets between subnets, which is Layer 3 behavior. Layer 2 switches forward on MAC addresses, while patch panels and media converters are passive or physical-layer components.

Question 9Data Center and Networking Fundamentals

An administrator needs to interconnect two data center switches so that all links actively carry traffic while remaining loop-free. Which technology should be used?

  1. Port mirroring
  2. STP with default priorities
  3. Link aggregation using LACP Correct answer.
  4. VLAN pruning

Explanation

Link aggregation bundles multiple physical links into one logical channel, so all member links forward traffic and the bundle is treated as a single path. Spanning Tree would block redundant links instead of using them, and pruning or mirroring do not aggregate bandwidth.

Question 10Data Center and Networking Fundamentals

A rack of servers must reach a storage array over a dedicated low-latency network that avoids TCP/IP stack overhead. Which technology best fits this requirement?

  1. Fibre Channel over Ethernet with FCoE
  2. Wi-Fi 6E
  3. InfiniBand Correct answer.
  4. 10GBASE-T Ethernet with TCP

Explanation

InfiniBand is a high-throughput, low-latency fabric designed for server-to-storage and cluster interconnect, with RDMA that bypasses much of the TCP/IP stack. FCoE still rides on Ethernet, and TCP and Wi-Fi add protocol overhead unsuitable for this requirement.

Question 11Data Center and Networking Fundamentals

In a spine-leaf architecture, what is the primary reason every leaf switch connects to every spine switch?

  1. To reduce the number of physical cabling runs
  2. To allow spanning tree to elect a single root
  3. To provide equal-cost multipath and consistent latency between any two leaves Correct answer.
  4. To centralize all VLAN gateways on one spine

Explanation

Full mesh leaf-to-spine connectivity gives each leaf an equal-cost path to every other leaf, so traffic is load-balanced across spines with predictable hop count and latency. It increases rather than reduces cabling, and the design relies on Layer 3 routing rather than spanning tree.

Question 12Data Center and Networking Fundamentals

Which statement correctly describes the difference between jumbo frames and standard Ethernet frames in a data center?

  1. Jumbo frames eliminate the need for TCP checksums
  2. Jumbo frames use an MTU larger than 1500 bytes and must be configured consistently end to end Correct answer.
  3. Jumbo frames are only supported on fiber media
  4. Jumbo frames reduce the maximum transmission unit to improve latency

Explanation

Jumbo frames raise the MTU above the standard 1500 bytes, typically to 9000, which reduces per-packet overhead for bulk transfers. Every device on the path, including switches and host NICs, must use a matching MTU or large packets will be dropped.

Question 13Data Center and Networking Fundamentals

A network engineer wants to carry multiple VLANs between two switches over a single physical link. What must be configured on that link?

  1. A port in a link aggregation group without tagging
  2. A trunk port using 802.1Q tagging Correct answer.
  3. A routed port with an IP address
  4. An access port assigned to one VLAN

Explanation

An 802.1Q trunk inserts a VLAN tag into each frame so multiple VLANs share one physical link. An access port carries a single untagged VLAN, and a routed port forwards between subnets rather than bridging multiple VLANs.

Question 14Data Center and Networking Fundamentals

Which mechanism allows a physical server to present multiple logical NICs, each with its own MAC address, to virtual machines?

  1. A virtual switch with virtual NICs Correct answer.
  2. NIC teaming
  3. Port mirroring
  4. A hardware load balancer

Explanation

A hypervisor virtual switch connects virtual NICs to the physical uplink, and each virtual NIC has its own MAC address so guests appear as distinct hosts on the network. Teaming bonds physical adapters, while load balancers and mirroring serve different functions.

Question 15Data Center and Networking Fundamentals

Which DNS record type maps a hostname to an IPv6 address?

  1. CNAME
  2. MX
  3. AAAA Correct answer.
  4. A

Explanation

An AAAA record holds a 128-bit IPv6 address for a hostname, while an A record holds a 32-bit IPv4 address. CNAME creates an alias to another name, and MX identifies mail exchangers.

Question 16Data Center and Networking Fundamentals

A data center uses BGP between its leaf switches and a route reflector. What is the main benefit of the route reflector in this design?

  1. It reduces the number of required iBGP peerings by reflecting routes to clients Correct answer.
  2. It converts BGP into a link-state protocol
  3. It encrypts all BGP updates
  4. It replaces the need for IP addressing

Explanation

A route reflector receives routes from clients and reflects them to other clients, replacing a full iBGP mesh with a hub-and-spoke peering model. BGP remains a path-vector protocol, and route reflection does not provide encryption or addressing.

Question 17Data Center and Networking Fundamentals

Which rack-level practice best mitigates hot spots in a data center with a hot-aisle/cold-aisle layout?

  1. Increasing the chilled water temperature setpoint
  2. Removing perforated floor tiles to reduce airflow
  3. Facing all server exhausts toward the cold aisle
  4. Installing blanking panels in unused rack units and sealing cable openings Correct answer.

Explanation

Blanking panels and sealed cable openings prevent cold air from bypassing equipment and stop hot exhaust from recirculating into the cold aisle. This preserves the intended front-to-back airflow path that the hot-aisle/cold-aisle layout depends on.

Question 18Data Center and Networking Fundamentals

What is the function of a PDU in a data center rack?

  1. It distributes electrical power to rack-mounted equipment Correct answer.
  2. It aggregates server network traffic
  3. It stores configuration backups
  4. It converts fiber optic signals to copper

Explanation

A power distribution unit takes input power and distributes it through multiple outlets to servers and switches in the rack, often with metering or remote switching. It is an electrical component and does not handle network traffic or signal conversion.

Question 19Data Center and Networking Fundamentals

A /30 point-to-point link is being configured between two routers. Why is /30 commonly chosen for this purpose?

  1. It provides exactly two usable addresses, minimizing address waste Correct answer.
  2. It supports broadcast traffic across multiple subnets
  3. It provides 62 usable addresses for future expansion
  4. It is the only mask supported on point-to-point links

Explanation

A /30 leaves 2 host bits, producing four addresses; after the network and broadcast addresses, exactly two remain for the endpoints. This conserves address space on links that only ever need two hosts.

Question 20Data Center and Networking Fundamentals

Which type of fiber is best suited for long-distance links between data center buildings several kilometers apart?

  1. Single-mode fiber Correct answer.
  2. OM3 multi-mode fiber
  3. Cat 6A twisted pair
  4. Direct-attach copper twinax

Explanation

Single-mode fiber has a small core that supports only a few propagation modes, which greatly reduces modal dispersion and allows links over many kilometers. Multi-mode and copper media are limited to much shorter distances.

Question 21Data Center and Networking Fundamentals

A storage array must be reachable by multiple servers, each seeing the same block device, over a switched fabric. Which technology natively provides this?

  1. iSCSI or Fibre Channel SAN Correct answer.
  2. HTTP object storage
  3. NFS exported from a file server
  4. SMB file shares

Explanation

SAN technologies such as iSCSI and Fibre Channel present block-level LUNs over a switched fabric so multiple hosts can access the same device with proper cluster-aware configuration. NFS, SMB, and HTTP object storage provide file or object access rather than block devices.

Question 22Data Center and Networking Fundamentals

In a leaf-spine fabric using Layer 3 routing to the host, what is the recommended purpose of the leaf switch's uplinks?

  1. To form a spanning tree topology with the spines
  2. To run routed links to every spine and use ECMP for load balancing Correct answer.
  3. To carry only management VLAN traffic
  4. To serve as a backup path that stays blocked

Explanation

Each leaf runs a routed link to every spine, and equal-cost multipath distributes flows across all spines simultaneously. This removes dependence on spanning tree and keeps all uplinks active rather than blocked.

Question 23Data Center and Networking Fundamentals

A server is configured with two PSUs on separate feeds, but a rack-level transfer switch feeds both from a single UPS. What risk remains?

  1. The server will draw more power than the circuit can supply
  2. The server will experience increased network latency
  3. The server will fail over to battery power immediately
  4. The server loses redundancy because both feeds share the UPS as a common failure point Correct answer.

Explanation

Dual-corded redundancy only helps if the two paths are independent. A shared UPS or transfer switch becomes a common point of failure, so a fault there removes power from both PSUs at once.

Question 24Data Center and Networking Fundamentals

A network team observes that large file transfers between two hosts stall while small requests succeed. Which cause is most consistent with this symptom?

  1. A DNS record pointing to the wrong AAAA address
  2. A duplicate MAC address on the same VLAN
  3. A spanning tree topology change
  4. An MTU mismatch along the path causing large packets to be dropped Correct answer.

Explanation

When an MTU mismatch exists, small packets pass but large packets that exceed the smallest MTU on the path are dropped, often with path MTU discovery blocked by filtering. The result is that bulk transfers stall while small transactions complete.

Question 25Data Center and Networking Fundamentals

A design requires that loss of a single spine switch reduce total fabric bandwidth but not connectivity between any two leaves. Which property must the design satisfy?

  1. Each leaf must connect to at least two spines Correct answer.
  2. Spines must be stacked into one logical switch
  3. All leaves must share a single VLAN
  4. Each leaf must connect to exactly one spine

Explanation

If every leaf connects to at least two spines, losing one spine leaves at least one path from each leaf to the rest of the fabric, so connectivity survives while aggregate bandwidth drops. Connecting to only one spine would isolate that leaf when its spine fails.

Question 26Data Center and Networking Fundamentals

Which RAID level provides block-level striping with distributed parity across all member disks, allowing the array to survive the failure of any single drive?

  1. RAID 5 Correct answer.
  2. RAID 1
  3. RAID 6
  4. RAID 10

Explanation

RAID 5 stripes data and parity blocks across every member disk, so parity is distributed rather than concentrated on a dedicated drive; any one disk can fail and be rebuilt from the surviving parity. RAID 6 adds a second parity block, while RAID 1 and RAID 10 rely on mirroring instead of parity.

Question 27Data Center and Networking Fundamentals

A server must resolve the hostname of a management controller to an IPv4 address on the data center network. Which DNS record type is queried for this purpose?

  1. CNAME
  2. AAAA
  3. PTR
  4. A Correct answer.

Explanation

An A record maps a hostname to a 32-bit IPv4 address, which is exactly what an IPv4 hostname lookup returns. AAAA records serve IPv6, CNAME records alias one name to another name, and PTR records answer reverse lookups from address back to name.

Question 28Data Center and Networking Fundamentals

A rack contains 20 servers drawing roughly 450 W each at peak. Which calculation correctly estimates the steady-state heat load the cooling system must remove from that rack?

  1. About 4.5 kW, because only half of the input power becomes heat
  2. About 9 kW, because electrical power consumed by the servers is converted almost entirely into heat Correct answer.
  3. About 0.9 kW, because efficiency losses dominate the load
  4. About 90 kW, because each server dissipates ten times its rated draw

Explanation

Nearly all electrical energy delivered to computing equipment is released as heat, so the cooling load in watts equals the IT power draw: 20 servers x 450 W = 9000 W, or about 9 kW. That figure is then used to size CRAC/CRAH capacity and airflow for the rack.

Question 29Data Center and Networking Fundamentals

An engineer must connect a new server to the data center LAN with the smallest possible broadcast domain while still allowing communication with hosts in other subnets. Which configuration meets this requirement?

  1. Place the server in the management VLAN so all devices share one broadcast domain
  2. Assign the server two addresses from the same subnet on the same NIC
  3. Place the server in a dedicated VLAN and route between VLANs with a Layer 3 gateway Correct answer.
  4. Configure the server port as a trunk carrying every VLAN in the fabric

Explanation

A VLAN defines a broadcast domain, so putting the server in its own VLAN limits broadcast traffic to that segment. A Layer 3 gateway, such as an SVI or router subinterface, then forwards traffic between that VLAN and other subnets.

Question 30Data Center and Networking Fundamentals

A fabric uses an overlay in which original Ethernet frames from servers are encapsulated inside IP packets between leaf switches. Which statement best describes the role of the underlay in this design?

  1. It carries only storage traffic, while the overlay forwards all server-to-server traffic
  2. It provides an IP-routed transport that delivers encapsulated packets between tunnel endpoints Correct answer.
  3. It must run STP to block redundant paths between the leaf switches
  4. It assigns MAC addresses to virtual machines on behalf of the hypervisors

Explanation

In an overlay such as VXLAN, the underlay is a routed IP network whose only job is to deliver encapsulated packets between tunnel endpoints, typically leaf switches acting as VTEPs. Because the underlay is Layer 3 and relies on equal-cost multipathing, it does not need a loop-free Layer 2 topology maintained by STP.

Question 31Storage Architecture and Design

In a Nutanix cluster, which component is responsible for maintaining the metadata that maps vDisk extents to physical storage and for serving I/O to the CVMs?

  1. The Cassandra distributed metadata store coordinated by the Stargate service Correct answer.
  2. The Acropolis File Services agent on each host
  3. The Prism Central virtual machine running on the cluster
  4. The hypervisor kernel module that bypasses the CVM

Explanation

Nutanix uses a distributed metadata ring built on Cassandra, with Stargate acting as the I/O and metadata service that routes vDisk requests to the owning CVM. This design lets any CVM locate data without a central lookup point. Prism Central is management, not the data path.

Question 32Storage Architecture and Design

A customer wants to protect against the loss of an entire Nutanix node without using replication factor 3. Which configuration achieves this?

  1. Replication factor 1 with frequent snapshots to a remote site
  2. Erasure coding stripe width 4 on a single node
  3. Replication factor 2 with both copies placed on the same node
  4. Replication factor 2 with data and metadata spread across two distinct nodes Correct answer.

Explanation

Replication factor 2 places two copies of data and metadata on distinct nodes, so a single node failure still leaves one accessible copy. Placing both copies on the same node defeats the purpose. Snapshots are point-in-time copies, not continuous redundancy.

Question 33Storage Architecture and Design

Which Nutanix capability allows a storage container to present different advertised capacities and reservations to different clients?

  1. Volume group load balancing
  2. Advertised capacity and reserved capacity settings on the container Correct answer.
  3. Storage container QoS policy
  4. Container replication factor override

Explanation

A storage container exposes advertised capacity, which is what clients see, separately from reserved capacity, which guarantees space. This lets administrators overcommit or reserve space per container. QoS throttles IOPS or bandwidth rather than capacity presentation.

Question 34Storage Architecture and Design

Which Nutanix feature provides thin-provisioned block storage to external physical hosts that cannot run a guest VM agent?

  1. Nutanix Files
  2. Nutanix Volumes Correct answer.
  3. Nutanix Objects
  4. Nutanix Mine

Explanation

Nutanix Volumes exposes iSCSI block targets backed by vDisks so external hosts can consume cluster storage without a guest agent. Files provides SMB/NFS shares, Objects provides S3-compatible object storage, and Mine is backup integration.

Question 35Storage Architecture and Design

What is the primary purpose of a protection domain in a Nutanix cluster?

  1. To assign storage QoS limits to individual vDisks
  2. To set the erasure coding stripe width for cold data
  3. To configure the replication factor of a storage container
  4. To define a set of VMs whose snapshots are coordinated for consistency Correct answer.

Explanation

A protection domain groups VMs so their snapshots are taken together, producing crash-consistent or application-consistent recovery points across the set. Replication factor, erasure coding, and QoS are configured elsewhere. This grouping is central to local and remote snapshot scheduling.

Question 36Storage Architecture and Design

Which statement best describes how Nutanix Distributed Storage Fabric handles a write when replication factor 2 is configured?

  1. The CVM writes all copies to a single node and rebalances nightly
  2. The CVM writes data only to the remote site and keeps metadata locally
  3. The CVM writes one copy locally and acknowledges immediately, replicating lazily in the background
  4. The CVM writes one copy locally and forwards the second copy to a remote CVM before acknowledging Correct answer.

Explanation

With RF2 the op-log and data are written locally and mirrored to a remote CVM, and the write is acknowledged only after the remote copy is durable. This ensures a node loss does not lose acknowledged writes. Lazy background replication would create a data-loss window.

Question 37Storage Architecture and Design

An administrator enables erasure coding on a storage container holding mostly cold data. What is the expected trade-off?

  1. Immediate conversion of existing hot data to a 4+2 stripe
  2. Lower capacity overhead but higher CPU and rebuild cost on reads and writes Correct answer.
  3. Higher capacity overhead but lower latency for all workloads
  4. Elimination of the need for replication factor entirely

Explanation

Erasure coding stores data as parity stripes, reducing raw capacity needed compared with replication, at the cost of more CPU work and slower rebuilds. It suits colder data where latency is less critical. New writes are encoded; existing data is converted gradually in the background.

Question 38Storage Architecture and Design

Which setting on a Nutanix volume group determines which CVM owns and serves the group's vDisks?

  1. The Prism Central category assignment
  2. The storage container replication factor
  3. The preferred leader or owning CVM assignment Correct answer.
  4. The protection domain schedule

Explanation

A volume group has an owning CVM that serves its vDisks, and administrators can set the preferred leader to control placement. Replication factor, protection domain schedules, and categories do not determine the serving CVM. This matters for latency tuning and for hosts using iSCSI.

Question 39Storage Architecture and Design

A cluster runs out of capacity on one node while others have free space. Which Nutanix mechanism restores balance?

  1. The Stargate service rewriting all vDisks to a single node
  2. The Acropolis scheduler migrating VMs
  3. The Prism Central capacity forecast job
  4. The Curator background service rebalancing extents across nodes Correct answer.

Explanation

Curator is the background scanner and rebalancer that moves extents to restore even utilization across nodes. Stargate serves I/O but does not perform the rebalancing policy. VM migration changes compute placement, not the distribution of stored extents.

Question 40Storage Architecture and Design

Which Nutanix storage feature is designed to provide S3-compatible object storage on the cluster?

  1. Nutanix Flow
  2. Nutanix Files
  3. Nutanix Volumes
  4. Nutanix Objects Correct answer.

Explanation

Nutanix Objects provides an S3-compatible object store that runs on the cluster and can scale out. Volumes provides iSCSI block targets, Files provides SMB/NFS, and Flow is network security policy. Object storage is the correct match for S3 APIs.

Question 41Storage Architecture and Design

How does Nutanix handle metadata resilience differently from data resilience in a cluster?

  1. Metadata uses erasure coding while data uses replication
  2. Metadata uses a Cassandra ring with multiple replicas independent of the container replication factor Correct answer.
  3. Metadata is replicated only during snapshots
  4. Metadata is stored only on the Prism leader CVM

Explanation

The distributed metadata ring keeps multiple copies of metadata across CVMs so any CVM can locate data, independent of a container's replication factor. This separation lets metadata remain highly available even for RF2 containers. Metadata is not confined to a single leader CVM.

Question 42Storage Architecture and Design

A VM's vDisk is configured with a storage QoS policy limiting IOPS. Where is this limit enforced in the Nutanix data path?

  1. At the CVM serving the vDisk through the Stargate I/O path Correct answer.
  2. At the guest operating system's block driver
  3. At the Prism Central REST API layer
  4. At the physical disk firmware

Explanation

Storage QoS policies are enforced by the CVM in the I/O path, throttling requests to the vDisk or volume group. The guest OS and Prism Central are not enforcement points, and disk firmware has no knowledge of per-vDisk policy. Enforcement in the CVM keeps limits consistent regardless of client.

Question 43Storage Architecture and Design

Which statement correctly contrasts a storage pool with a storage container in Nutanix?

  1. A storage pool is a snapshot schedule, while a container is a replication topology
  2. A storage pool is S3 object storage, while a container is iSCSI block storage
  3. A container aggregates physical disks, while a pool presents logical storage with policies
  4. A storage pool aggregates physical disks, while a container presents logical storage with its own policies Correct answer.

Explanation

Storage pools group physical devices into a fault-tolerant unit, and containers sit above pools to present logical storage with settings such as replication factor and compression. Containers are the units administrators assign to VMs. Snapshots, replication, and protocols are separate concepts.

Question 44Storage Architecture and Design

An administrator needs point-in-time recovery of a VM with minimal capacity consumption across many snapshots. Which Nutanix capability supports this?

  1. Log shipping to a remote site
  2. Full clones of the vDisk for each snapshot
  3. Erasure coding of the snapshot chain
  4. Redirect-on-write snapshots that share unchanged extents Correct answer.

Explanation

Nutanix snapshots are metadata operations that reference existing extents, so unchanged data is shared rather than duplicated. This keeps many recovery points space-efficient. Full clones copy data and consume capacity per copy, and log shipping is a replication technique.

Question 45Storage Architecture and Design

What role does the op-log play in the Nutanix write path?

  1. It encrypts data at rest on each node
  2. It stores only snapshot metadata for protection domains
  3. It caches reads to reduce latency on hot data
  4. It provides a persistent, replicated staging area for writes before they are destaged to extent groups Correct answer.

Explanation

Incoming writes land in the op-log, which is replicated for durability, and are later destaged into extent groups. This absorbs bursts and preserves ordering. Snapshot metadata, read caching, and encryption are handled by other components.

Question 46Storage Architecture and Design

A customer wants the cluster to tolerate the simultaneous loss of two nodes while keeping capacity overhead reasonable. Which approach is most appropriate?

  1. Use replication factor 2 across all containers
  2. Use replication factor 1 with remote snapshots
  3. Use erasure coding with a stripe that tolerates two failures, such as 4+2 Correct answer.
  4. Use a single storage pool with no redundancy

Explanation

A 4+2 erasure coding stripe can reconstruct data after two component failures while using less raw capacity than three full replicas. RF2 tolerates only one failure, and RF1 or no redundancy tolerates none. The stripe's parity count determines how many failures are survivable.

Question 47Storage Architecture and Design

In a Nutanix cluster, why can a VM continue running briefly if its owning CVM fails?

  1. Prism Central takes over I/O for the failed CVM
  2. The hypervisor has a full local copy of every vDisk
  3. The physical disk controller serves I/O directly to the guest
  4. The vDisk is remapped to another CVM that can access the replicated data and metadata Correct answer.

Explanation

Because data and metadata are replicated and the metadata ring is distributed, another CVM can assume ownership and serve the vDisk after a CVM failure. This remapping is why VM availability does not depend on a single CVM. The hypervisor does not keep full local copies.

Question 48Storage Architecture and Design

Which consideration most affects the usable capacity reported for a container using RF2 with erasure coding disabled?

  1. Roughly half of raw capacity is usable because each extent is stored twice Correct answer.
  2. Usable capacity depends only on the number of VMs in the container
  3. Usable capacity equals raw capacity because metadata is not counted
  4. Usable capacity is one quarter of raw capacity due to parity

Explanation

With RF2 each extent has two copies, so usable capacity is about half of raw after overheads. Metadata and reserved space further reduce the figure slightly. Parity overhead applies to erasure coding, not RF2.

Question 49Storage Architecture and Design

An administrator observes high read latency on a container while CPU and network are healthy. Which Nutanix mechanism is most likely responsible for promoting frequently accessed extents?

  1. The volume group iSCSI initiator
  2. The Prism Central report generator
  3. The protection domain snapshot scheduler
  4. The extent cache and tiering behavior managed by the CVM Correct answer.

Explanation

CVMs maintain caches and manage data placement so hot extents are served efficiently; cache behavior strongly influences read latency. Snapshot scheduling, reporting, and iSCSI initiators do not promote hot extents. Investigating cache and tiering is the right first step.

Question 50Storage Architecture and Design

Which Nutanix feature lets an administrator apply different replication factors to different workloads on the same cluster?

  1. Setting a single cluster-wide replication factor only
  2. Enabling erasure coding on the cluster to override replication
  3. Assigning categories to VMs to change their replication factor
  4. Configuring separate storage containers with distinct replication settings Correct answer.

Explanation

Replication factor is set per storage container, so multiple containers with different settings let workloads on the same cluster have different redundancy. There is no single mandatory cluster-wide value. Categories do not change replication factor, and erasure coding does not override it.