I85PREP
⌘ K
NSEA study materials
NSEA study resources

NSEA practice questions

One question at a time. Build your NSEA 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 1Nutanix Fundamentals and Hyperconverged Infrastructure Concepts

In a Nutanix hyperconverged infrastructure cluster, which component runs directly on each node and provides the distributed storage layer that pools local disks across the cluster?

  1. The Controller VM (CVM) Correct answer.
  2. The Prism Central virtual machine
  3. The cluster's external SAN gateway
  4. The Acropolis hypervisor agent

Explanation

Each node in a Nutanix cluster hosts a Controller VM that owns the local disks and participates in the distributed storage fabric, so disks from all nodes are presented as one pool. Prism Central is a management plane, not the storage engine, and HCI clusters do not rely on an external SAN.

Question 2Nutanix Fundamentals and Hyperconverged Infrastructure Concepts

Which statement best describes what makes an architecture 'hyperconverged' rather than merely 'converged'?

  1. All workloads run on a single large host that is partitioned into logical segments
  2. Compute and storage are separate tiers connected by a dedicated storage network and scaled independently
  3. Storage is virtualized in software but still resides on a shared external array
  4. Compute and storage are combined in the same nodes and pooled by software, scaling out by adding nodes Correct answer.

Explanation

Hyperconvergence collapses compute and storage into the same node and uses distributed software to pool those local resources, so capacity and performance grow by adding nodes. Converged systems keep separate compute and storage tiers, which is what the other options describe.

Question 3Nutanix Fundamentals and Hyperconverged Infrastructure Concepts

A Nutanix cluster uses replication factor 2. How many copies of each data extent does the distributed storage fabric maintain, and what does this imply for usable capacity?

  1. One copy plus parity, so usable capacity is close to raw capacity
  2. Three copies, so roughly one third of raw capacity is usable
  3. Two copies, so roughly half of raw disk capacity is usable before other overheads Correct answer.
  4. Two copies, but usable capacity equals raw capacity because copies share space

Explanation

Replication factor 2 keeps two copies of each extent on different nodes or failure domains, so the data is protected against a single component failure at the cost of roughly half the raw capacity. Parity-based schemes, not RF2, achieve capacity efficiency closer to raw.

Question 4Nutanix Fundamentals and Hyperconverged Infrastructure Concepts

What is the primary role of Prism Central in a Nutanix environment?

  1. It provides centralized management, monitoring, and policy across multiple registered clusters Correct answer.
  2. It replaces the Controller VM as the local storage engine on each node
  3. It is the physical switch fabric that connects nodes in a cluster
  4. It is the hypervisor that runs guest virtual machines on each host

Explanation

Prism Central is a separate management appliance that registers multiple clusters and offers centralized visibility, categories, and policy-driven operations. It does not perform per-node storage I/O, run the hypervisor, or provide the physical network.

Question 5Nutanix Fundamentals and Hyperconverged Infrastructure Concepts

A cluster administrator wants a single management interface to oversee several independent Nutanix clusters located in different data centers. Which approach fits?

  1. Replace each Controller VM with a Prism Central VM
  2. Convert all clusters into one Prism Element domain
  3. Register each cluster with a Prism Central instance and manage them from there Correct answer.
  4. Log in to each cluster's Prism Element and aggregate reports manually

Explanation

Prism Central is designed to manage multiple clusters by registering them, giving one pane of glass across sites. Prism Element manages a single cluster, and Controller VMs are storage components that cannot be swapped for management appliances.

Question 6Nutanix Fundamentals and Hyperconverged Infrastructure Concepts

Which Nutanix feature allows a VM's storage to remain locally accessible on its host for reads while still being protected across the cluster?

  1. Deduplication, which removes duplicate blocks before reads
  2. Live migration, which moves the VM to the node holding the only copy
  3. Data locality, where the distributed storage fabric keeps a read-local copy near the VM Correct answer.
  4. Snapshots, which copy VM data to remote nodes on every read

Explanation

The distributed storage fabric maintains locality so a VM can read from local media when a copy is present, reducing network hops while replicas elsewhere preserve resilience. Snapshots, deduplication, and migration address different concerns and do not describe read locality.

Question 7Nutanix Fundamentals and Hyperconverged Infrastructure Concepts

Which Nutanix component provides the REST API and CLI used to script cluster operations such as creating VMs and managing storage containers?

  1. The Controller VM's iSCSI target only
  2. Prism, exposed through Prism Element and Prism Central APIs Correct answer.
  3. The hypervisor's kernel debugger
  4. The baseboard management controller

Explanation

Nutanix exposes management operations through Prism APIs, available at both the cluster level (Prism Element) and the multi-cluster level (Prism Central). The other choices are low-level or unrelated interfaces that do not offer the documented automation API.

Question 8Nutanix Fundamentals and Hyperconverged Infrastructure Concepts

What does a storage container represent in a Nutanix cluster?

  1. A logical policy boundary that defines settings such as replication factor and compression for the VMs placed in it Correct answer.
  2. A snapshot schedule applied to the Controller VM itself
  3. A physical disk shelf attached to a single node
  4. A VLAN that isolates storage traffic between nodes

Explanation

A storage container is a logical construct that carries data policies like replication factor, compression, and deduplication, and VMs' virtual disks are assigned to it. It is not physical hardware, a network segment, or a snapshot schedule.

Question 9Nutanix Fundamentals and Hyperconverged Infrastructure Concepts

A node in a four-node cluster loses power entirely. What happens to the data that was stored on that node if replication factor 2 was in use?

  1. The Controller VM on each remaining node shuts down to avoid split brain
  2. All data on that node is permanently lost because only one copy existed
  3. The surviving replica on another node continues to serve I/O, and the cluster rebuilds redundancy when the node returns Correct answer.
  4. The cluster halts all I/O until the failed node is replaced

Explanation

With RF2, each extent has a second copy on a different failure domain, so the cluster keeps serving reads and writes from surviving replicas and re-protects data after the node recovers. Losing one node does not stop the cluster or destroy the data.

Question 10Nutanix Fundamentals and Hyperconverged Infrastructure Concepts

Which statement correctly distinguishes a Nutanix 'node' from a Nutanix 'block'?

  1. A node is the Prism Central appliance, while a block is a cluster of nodes
  2. A node is a single server with compute and local disks, while a block is a chassis that houses multiple nodes Correct answer.
  3. A node is a network switch, while a block is a storage array
  4. A node is a logical container, while a block is a single disk drive

Explanation

In Nutanix hardware terminology, a node is an individual server contributing CPU, memory, and disks, and a block is the chassis that contains several nodes plus shared components such as power and networking. The two terms describe different physical levels.

Question 11Nutanix Fundamentals and Hyperconverged Infrastructure Concepts

Which capability is provided by Nutanix Acropolis in a cluster running AHV?

  1. A guest operating system installed inside each virtual machine
  2. A backup appliance that replaces snapshots
  3. A native hypervisor with integrated VM management and distributed storage services Correct answer.
  4. A hardware RAID controller firmware

Explanation

Acropolis covers the AHV hypervisor and the management services that create and operate VMs, tightly integrated with the distributed storage fabric. It is not a guest OS, firmware, or a backup appliance.

Question 12Nutanix Fundamentals and Hyperconverged Infrastructure Concepts

An administrator needs to ensure that two replicas of the same data are never placed on nodes that share the same power source or rack. Which concept addresses this?

  1. Deduplication, which removes duplicate blocks
  2. Storage containers, which set the VM's guest OS type
  3. Failure domains, which define the boundaries used when placing replicas Correct answer.
  4. Prism Central categories, which tag VMs for reporting

Explanation

Failure domains describe the physical or logical boundaries, such as a block or rack, that the storage fabric considers when placing replicas so that a single failure cannot take out all copies. Containers, deduplication, and categories serve different purposes.

Question 13Nutanix Fundamentals and Hyperconverged Infrastructure Concepts

What is the effect of enabling compression on a Nutanix storage container?

  1. Data is moved to an external array to free local capacity
  2. Data is encrypted with a cluster key before being written to disk
  3. Data is compressed before being written to disk, trading some CPU for reduced storage consumption Correct answer.
  4. Data is replicated to an additional site for disaster recovery

Explanation

Container-level compression reduces the bytes written to local media by compressing data inline, which saves capacity at the cost of some processing overhead. Encryption, remote replication, and tiering to external arrays are separate features.

Question 14Nutanix Fundamentals and Hyperconverged Infrastructure Concepts

Which Nutanix capability allows a VM to continue running on a surviving node when its current host fails, without manual intervention?

  1. High availability, where the cluster restarts the VM on another node Correct answer.
  2. Data locality, where reads are served from the nearest copy
  3. Thin provisioning, where disk space is allocated on demand
  4. Deduplication, where duplicate blocks are collapsed across nodes

Explanation

Nutanix high availability detects host failure and restarts affected VMs on other nodes in the cluster so services recover automatically. Deduplication, thin provisioning, and locality affect storage efficiency or performance, not VM restart.

Question 15Nutanix Fundamentals and Hyperconverged Infrastructure Concepts

A workload requires very high random read throughput on a single VM. Which Nutanix design choice most directly supports low-latency reads for that VM?

  1. Keeping a read-local replica on the host running the VM so reads avoid network hops Correct answer.
  2. Increasing the replication factor to 3 so more copies exist cluster-wide
  3. Moving the VM's disks to an external SAN
  4. Disabling compression so every block is stored uncompressed

Explanation

When a copy of the data is local to the host running the VM, reads are served from local media and avoid network latency, which is the main lever for read performance. Raising replication factor or disabling compression does not specifically optimize read locality, and external SAN contradicts HCI.

Question 16Nutanix Fundamentals and Hyperconverged Infrastructure Concepts

What does Nutanix Prism Element manage?

  1. Multiple clusters across data centers from one appliance
  2. Only the physical network switches in a rack
  3. Only the guest operating systems inside VMs
  4. A single Nutanix cluster, including its hosts, storage, and VMs Correct answer.

Explanation

Prism Element is the per-cluster management interface for hosts, storage, and VMs in that cluster, while multi-cluster oversight is the job of Prism Central. It does not manage physical switches or guest OS internals.

Question 17Nutanix Fundamentals and Hyperconverged Infrastructure Concepts

Which statement about scaling a Nutanix cluster is accurate?

  1. Adding a node adds both compute and storage capacity in a single step Correct answer.
  2. Adding a node adds only storage capacity while compute stays fixed
  3. Scaling requires shutting down the cluster to add capacity
  4. Scaling requires adding a separate storage array before adding compute

Explanation

Because compute and storage are combined per node, adding a node increases CPU, memory, and disk capacity together, which is the scale-out promise of HCI. Separate storage arrays and full-cluster shutdowns are not required.

Question 18Nutanix Fundamentals and Hyperconverged Infrastructure Concepts

An administrator wants point-in-time recoverable copies of a VM's disks taken on a schedule with minimal capacity impact. Which Nutanix mechanism is intended for this?

  1. Prism Central role-based access control
  2. Protection domains with snapshot schedules Correct answer.
  3. Storage container compression settings
  4. The hypervisor's live migration feature

Explanation

Protection domains group VMs and apply snapshot schedules, producing point-in-time copies that are space-efficient and can be replicated. Compression, access control, and live migration do not create scheduled recovery points.

Question 19Nutanix Fundamentals and Hyperconverged Infrastructure Concepts

Why does the Nutanix distributed storage fabric use checksums on data?

  1. To compress each block before writing it to disk
  2. To encrypt data at rest with a cluster key
  3. To route storage traffic between VLANs
  4. To detect silent corruption so that a good replica can be used and the bad copy repaired Correct answer.

Explanation

Checksums let the storage fabric verify data integrity on reads; if a copy is corrupt, another replica can serve the request and the damaged copy is repaired. Compression, encryption, and network routing are handled by other features.

Question 20Nutanix Fundamentals and Hyperconverged Infrastructure Concepts

A cluster has four nodes and uses replication factor 3. What is the minimum number of node failures the data can survive while still retaining at least one complete copy of every extent?

  1. Zero failures
  2. Four failures
  3. Three failures
  4. Two failures Correct answer.

Explanation

With three replicas placed on distinct failure domains, losing two nodes still leaves one copy of each extent available, though the cluster operates in a degraded state until it re-protects. Losing three would remove all copies.

Question 21Nutanix Fundamentals and Hyperconverged Infrastructure Concepts

Which Nutanix feature lets an administrator tag VMs with business metadata that can then drive policy and reporting across clusters?

  1. Controller VM hostnames
  2. iSCSI initiator IQNs
  3. Categories in Prism Central Correct answer.
  4. Storage container names

Explanation

Prism Central categories attach key-value metadata to VMs and other objects, enabling policy application and reporting across registered clusters. Container names, CVM hostnames, and IQNs are technical identifiers, not business metadata.

Question 22Nutanix Fundamentals and Hyperconverged Infrastructure Concepts

An administrator observes that a VM's writes are slower than expected and finds the VM is writing to a storage container with replication factor 3 on a small cluster. What is the most likely explanation?

  1. Each write must be committed to multiple replicas before it is acknowledged, adding latency Correct answer.
  2. Deduplication is disabled, so every write is stored twice
  3. Prism Central is polling the cluster too frequently
  4. The Controller VM is compressing data, which doubles write size

Explanation

Higher replication factors increase write amplification because a write is acknowledged only after the required replicas are committed, which can raise latency on small clusters. Deduplication, compression, and management polling do not explain the multi-replica write path.

Question 23Nutanix Fundamentals and Hyperconverged Infrastructure Concepts

What is the purpose of the Nutanix Cluster Check (NCC) health tool?

  1. It replaces Prism Central for multi-cluster management
  2. It encrypts data at rest on each node
  3. It provisions new nodes into the cluster automatically
  4. It runs automated checks and surfaces findings about cluster health and configuration Correct answer.

Explanation

NCC performs automated health and configuration checks and reports findings that help administrators prevent or diagnose issues. It does not expand the cluster, replace Prism Central, or provide encryption.

Question 24Nutanix Fundamentals and Hyperconverged Infrastructure Concepts

Which statement best explains why HCI clusters can improve operational simplicity compared with three-tier architectures?

  1. A single vendor stack and software-defined storage remove separate storage array and SAN management Correct answer.
  2. Storage and compute are administered by different specialized teams by design
  3. Each workload must be manually pinned to a dedicated storage LUN
  4. Capacity is added only by purchasing larger disks for existing nodes

Explanation

Because compute and storage are integrated and managed through one software stack, teams avoid separate array, SAN fabric, and LUN administration, reducing operational overhead. Manual LUN pinning, split administration, and disk-only upgrades describe the opposite.

Question 25Nutanix Fundamentals and Hyperconverged Infrastructure Concepts

In a Nutanix cluster, where does a guest VM's virtual disk actually reside?

  1. As vDisk extents managed by the distributed storage fabric across the nodes' local disks Correct answer.
  2. As a partition on the hypervisor's USB boot device
  3. As a file on the Prism Central appliance's boot disk
  4. As an iSCSI LUN on an external array

Explanation

Virtual disks are broken into extents that the distributed storage fabric places on local disks across nodes according to container policy. They are not stored on the Prism Central appliance, an external array, or the boot device.

Question 26Nutanix Fundamentals and Hyperconverged Infrastructure Concepts

In a Nutanix cluster, which network is dedicated to carrying storage replication traffic between Controller VMs on different nodes?

  1. The backplane network, which is isolated from the management and guest VM networks Correct answer.
  2. The out-of-band IPMI network used for hardware power control
  3. The management network used by Prism Element and Prism Central
  4. The guest VM network that carries tenant application traffic

Explanation

A Nutanix cluster separates traffic into distinct networks: management, guest VM, and backplane. The backplane is a dedicated, often internal, network that carries replicated write data between Controller VMs so that storage traffic does not compete with tenant or management traffic.

Question 27Nutanix Fundamentals and Hyperconverged Infrastructure Concepts

An administrator wants a Nutanix cluster to tolerate the loss of two nodes while still keeping a full copy of every write. Which replication factor setting is required?

  1. Replication factor 2
  2. Replication factor 3 Correct answer.
  3. Replication factor 1
  4. Replication factor 0

Explanation

Replication factor defines how many copies of each write exist across failure domains. With RF3, three copies exist on three separate nodes, so two nodes can fail and one copy still remains; RF2 only survives a single node failure.

Question 28Nutanix Fundamentals and Hyperconverged Infrastructure Concepts

A cluster uses RF2 across four nodes. An administrator adds two more nodes to the same cluster. What is the primary effect on resilience and capacity distribution?

  1. The replication factor automatically increases to match the larger node count
  2. Existing data stays pinned to the original four nodes until a manual migration is triggered
  3. Capacity per node doubles because the cluster redistributes parity instead of replicas
  4. Rebalancing spreads data across the new nodes, and more failure domains become available for placement Correct answer.

Explanation

Adding nodes expands the pool of failure domains, and the distributed storage fabric rebalances existing data across the new nodes so capacity and load are spread evenly. RF remains whatever the administrator configured; it does not change on its own.

Question 29Nutanix Fundamentals and Hyperconverged Infrastructure Concepts

An administrator needs to protect VMs against the loss of an entire physical site while keeping a single management view. Which Nutanix capability addresses this requirement?

  1. Protection domains that replicate selected VMs to a remote cluster Correct answer.
  2. Storage containers configured with RF3 on the local cluster
  3. Prism Element dashboards aggregated per cluster
  4. Volume groups presented over iSCSI to external hosts

Explanation

Protection domains define a set of VMs and a schedule for replicating them to a remote Nutanix cluster, which provides site-level disaster recovery. Prism Central can then manage both clusters from one console, giving the single management view.

Question 30Nutanix Fundamentals and Hyperconverged Infrastructure Concepts

A three-node cluster uses RF3. One node is taken offline for maintenance, and the administrator then applies a policy that requires RF3 on the same cluster. What is the most accurate outcome?

  1. The cluster silently drops to RF2 and continues writing normally without any alert
  2. The cluster automatically shrinks the vDisk extent size to compensate for the missing node
  3. The cluster promotes the remaining two nodes to RF3 by storing two replicas on each
  4. New writes can still achieve three copies only after the node returns, so the cluster may block or degrade writes while RF3 cannot be satisfied Correct answer.

Explanation

RF3 requires three independent failure domains, so with only two nodes available the policy cannot be met. The cluster will not silently rewrite the policy; writes may be blocked or degraded, and health alerts surface the unmet requirement until the node returns.

Question 31Core Platform Architecture and Distributed System Design

In a distributed routing architecture, what is the primary purpose of separating the control plane from the data plane?

  1. To ensure that all traffic follows a single deterministic path
  2. To guarantee that every packet is encrypted before it leaves the device
  3. To eliminate the need for routing protocols between devices
  4. To allow forwarding decisions to be computed centrally or in software while packet handling stays on high-throughput hardware Correct answer.

Explanation

Control-plane and data-plane separation lets forwarding logic be computed by software controllers or centralized services while dedicated hardware performs high-speed packet forwarding. This improves programmability, policy consistency, and independent scaling of each function. Encryption and path determinism are unrelated concerns.

Question 32Core Platform Architecture and Distributed System Design

Which statement best describes the role of a distributed state database in a modern network operating system?

  1. It caches only transient packet counters local to each line card
  2. It stores packet payloads for later forensic analysis
  3. It stores the authoritative configuration and operational state that all cluster members can read consistently Correct answer.
  4. It replaces the need for any routing protocol adjacencies

Explanation

Modern network operating systems use a distributed database to hold configuration and operational state so every node in the cluster shares a consistent view. This underpins features such as nonstop routing and unified management. It is not a packet-payload store or a substitute for protocol adjacencies.

Question 33Core Platform Architecture and Distributed System Design

What does the term 'eventual consistency' mean when applied to a distributed network state store?

  1. Reads always return the most recent write from any node
  2. The database guarantees linearizable reads at all times
  3. Writes are always rejected until every replica acknowledges them
  4. Replicas may temporarily diverge but converge to the same value once updates propagate Correct answer.

Explanation

Eventual consistency allows replicas to serve reads while updates are still propagating, accepting temporary divergence in exchange for availability and partition tolerance. Convergence is guaranteed only after propagation completes. Strong or linearizable consistency is the opposite guarantee.

Question 34Core Platform Architecture and Distributed System Design

In a spine-leaf data center fabric, why is every leaf typically connected to every spine?

  1. To provide uniform, predictable any-to-any latency and equal-cost multipathing between servers Correct answer.
  2. To centralize all inter-subnet routing on a single spine
  3. To create a looped topology that requires STP to block redundant paths
  4. To reduce the total number of physical links required in the fabric

Explanation

A full-mesh leaf-to-spine topology gives consistent hop count between any two leaves, enabling ECMP and predictable east-west latency. Routing protocols such as BGP replace spanning tree in this design. It increases rather than reduces cabling, and routing is distributed across leaves.

Question 35Core Platform Architecture and Distributed System Design

Which mechanism allows a routing protocol to maintain forwarding state during a supervisor switchover on a modular chassis?

  1. Spanning tree root guard
  2. Graceful restart with helper support from neighbors Correct answer.
  3. DHCP snooping
  4. Port mirroring

Explanation

Graceful restart lets a restarting router preserve forwarding entries while its control plane recovers, with neighbors acting as helpers to retain routes. This avoids traffic disruption during supervisor failover. The other options are Layer 2 security or monitoring features unrelated to control-plane restart.

Question 36Core Platform Architecture and Distributed System Design

What is the function of a route reflector in a large-scale BGP deployment?

  1. It encrypts BGP sessions between autonomous systems
  2. It assigns IP addresses to BGP peers dynamically
  3. It reduces the number of required iBGP sessions by reflecting routes between clients Correct answer.
  4. It converts external BGP routes into interior gateway protocol routes automatically

Explanation

A route reflector removes the full-mesh iBGP requirement by relaying routes from one client to others within a cluster. This scales iBGP deployments significantly. It does not perform protocol conversion, encryption, or address assignment.

Question 37Core Platform Architecture and Distributed System Design

In a VXLAN overlay, what is the role of the VXLAN Network Identifier (VNI)?

  1. It identifies the physical underlay VLAN tag on the spine uplink
  2. It identifies a virtual Layer 2 segment carried inside the tunnel Correct answer.
  3. It encodes the source IP address of the tunnel endpoint
  4. It sets the MTU for the tunnel path

Explanation

The 24-bit VNI scopes a VXLAN segment, allowing many isolated Layer 2 domains to share a common IP underlay. It is carried in the VXLAN header, independent of any underlay VLAN. It does not encode endpoint addresses or MTU.

Question 38Core Platform Architecture and Distributed System Design

Which characteristic distinguishes a scale-out architecture from a scale-up architecture?

  1. Capacity is increased by adding more commodity nodes rather than by upgrading a single larger node Correct answer.
  2. Performance is improved solely through faster single-core CPUs
  3. Capacity is increased only by replacing the chassis with a higher-capacity model
  4. Redundancy is achieved only through redundant power supplies

Explanation

Scale-out adds more nodes to distribute load, while scale-up enlarges a single system. Scale-out designs favor commodity hardware, horizontal elasticity, and distributed failure domains. Chassis replacement, faster single cores, and power redundancy are not defining traits of scale-out.

Question 39Core Platform Architecture and Distributed System Design

A network operator wants configuration changes applied atomically across all nodes in a cluster. Which approach best supports this requirement?

  1. Commit the change through the distributed state database with a two-phase commit or transactional model Correct answer.
  2. Push changes only to the standby supervisor and let it sync later
  3. Apply the change node by node with manual verification after each step
  4. Rely on each node to poll for configuration files independently

Explanation

Transactional commit through the distributed state database ensures all nodes accept or reject the change together, preventing partial configuration states. Manual per-node application and independent polling create windows of inconsistency. Syncing only the standby leaves the active node unsynchronized.

Question 40Core Platform Architecture and Distributed System Design

A design uses a distributed hash table to place tenants across nodes. What happens when a node fails and consistent hashing with virtual nodes is in use?

  1. All keys are rehashed and every tenant is remapped across the cluster
  2. Only the keys mapped to the failed node's virtual nodes are redistributed to successor nodes Correct answer.
  3. The cluster stops accepting writes until the failed node returns
  4. Keys on the failed node are permanently lost

Explanation

Consistent hashing limits redistribution to the key ranges owned by the failed node's virtual nodes, which move to their successors. This minimizes data movement and disruption. Full rehashing, cluster-wide write blocking, and permanent loss are not inherent to the design, especially with replication.

Question 41Core Platform Architecture and Distributed System Design

An engineer observes that a distributed control plane is split-brained after a network partition. Which mechanism most directly prevents two halves from both acting as primary?

  1. Quorum-based leader election requiring a majority of cluster members Correct answer.
  2. Configuring static routes on each node
  3. Enabling jumbo frames on the cluster interconnect
  4. Increasing the hello interval on all cluster links

Explanation

Quorum-based election ensures only the partition containing a majority can elect a leader, preventing split-brain. Timing tweaks, MTU changes, and static routes do not establish mutual exclusion between partitions.

Question 42Core Platform Architecture and Distributed System Design

Which design choice best mitigates the risk of a single ToR switch failure causing application downtime in a rack?

  1. Disabling link-state tracking on server ports
  2. Dual-homing servers to two ToR switches with an MLAG or EVPN multihoming configuration Correct answer.
  3. Using a single uplink from the ToR to the spine
  4. Connecting all servers in the rack to one ToR and relying on RAID

Explanation

Dual-homing servers to redundant ToR switches with MLAG or EVPN multihoming removes the ToR as a single point of failure. RAID protects storage, not network paths. Single uplinks and disabled tracking increase exposure.

Question 43Core Platform Architecture and Distributed System Design

In an EVPN-VXLAN fabric, what is the purpose of the Route Type 2 advertisement?

  1. It carries the fabric's multicast group membership
  2. It advertises MAC and IP address bindings for hosts to enable efficient forwarding Correct answer.
  3. It signals the designated forwarder for a multihomed segment
  4. It advertises the underlay point-to-point link addresses

Explanation

EVPN Route Type 2 carries MAC/IP binding information, enabling control-plane learning and efficient host forwarding across the overlay. Type 1 handles Ethernet auto-discovery and Type 4 handles designated forwarder election. Type 3 carries inclusive multicast membership.

Question 44Core Platform Architecture and Distributed System Design

A team needs a fabric that supports Layer 2 extension across sites while keeping broadcast domains bounded. Which combination best meets this requirement?

  1. OSPF area 0 extended across all sites
  2. A flat Layer 2 trunk between sites with STP
  3. Static routes with proxy ARP
  4. EVPN with VXLAN data plane and selective stretch of only required VLANs Correct answer.

Explanation

EVPN-VXLAN provides controlled Layer 2 extension with a distributed control plane, and stretching only needed VLANs limits broadcast scope. Flat trunks invite STP instability and large failure domains. OSPF and static routing do not extend Layer 2 domains.

Question 45Core Platform Architecture and Distributed System Design

What is the primary benefit of an anycast service address in a distributed platform?

  1. It eliminates the need for health checking
  2. It forces all traffic through a single central gateway
  3. Clients reach the nearest instance automatically, and a failed instance is withdrawn from routing Correct answer.
  4. It guarantees that all requests from a client go to the same instance for its lifetime

Explanation

Anycast advertises the same address from multiple locations so routing delivers clients to the topologically nearest instance. Withdrawing the advertisement on failure redirects traffic. It does not guarantee session affinity, replace health checks, or centralize traffic.

Question 46Core Platform Architecture and Distributed System Design

An architect is choosing between a shared-memory and a message-passing model for inter-process communication in a distributed system. Which statement is accurate?

  1. Shared memory scales across data centers more easily than message passing
  2. Shared memory requires processes to be on the same host, while message passing works across hosts Correct answer.
  3. Message passing cannot carry structured data between processes
  4. Shared memory guarantees ordering across network partitions

Explanation

Shared memory is confined to processes on the same physical or virtual host, whereas message passing operates over networks and therefore spans hosts. Message passing can carry structured serialized data. Neither model alone guarantees ordering across partitions.

Question 47Core Platform Architecture and Distributed System Design

Which statement about the CAP theorem is correct for a network controller cluster?

  1. Partition tolerance can be dropped by adding more nodes
  2. CAP applies only to single-node databases
  3. During a partition, the system must choose between consistency and availability Correct answer.
  4. A system can guarantee consistency, availability, and partition tolerance simultaneously

Explanation

CAP states that under a network partition a distributed system must sacrifice either consistency or availability. Partition tolerance is unavoidable in real networks, so the practical trade-off is between the other two. Adding nodes does not eliminate partitions, and CAP applies to distributed systems generally.

Question 48Core Platform Architecture and Distributed System Design

What is the effect of enabling Bidirectional Forwarding Detection (BFD) on an OSPF adjacency?

  1. OSPF adjacencies are replaced by static routes
  2. BFD increases the OSPF hello interval to reduce overhead
  3. BFD encrypts OSPF hello packets
  4. Failures are detected in sub-second time, allowing faster reconvergence than hello timers alone Correct answer.

Explanation

BFD provides rapid liveness detection independent of routing protocol timers, so OSPF can reconverge much faster than hello-based detection allows. It does not replace adjacencies, encrypt hellos, or lengthen hello intervals.

Question 49Core Platform Architecture and Distributed System Design

A distributed platform requires that configuration writes survive the loss of any single node. Which replication strategy satisfies this?

  1. Synchronous replication to a quorum of nodes before acknowledging the write Correct answer.
  2. Write-back caching on the receiving node only
  3. Periodic snapshot export to remote storage once per day
  4. Asynchronous replication to a single backup node

Explanation

Acknowledging a write only after a quorum of replicas persists it ensures the data survives the loss of any single node. Single-backup async replication, local caching, and daily snapshots all leave a window where a node loss could lose recent writes.

Question 50Core Platform Architecture and Distributed System Design

Which factor most limits the practical scale of a single BGP route reflector cluster?

  1. The number of physical interfaces on the route reflector
  2. The maximum length of the BGP community attribute
  3. The CPU and memory required to process and store routes for all clients Correct answer.
  4. The TTL value configured on BGP sessions

Explanation

Route reflectors concentrate route processing and storage for their clients, so CPU and memory become the scaling bottleneck. Interface count, community length, and TTL are not the primary limits on cluster scale.