10 free sample questions from a bank of 487, with the correct answers and explanations. No signup required — start practising right now.
1Refer to the exhibit. An engineer is deploying a Cisco ACI environment but experiences a STP loop between SW1 and SW2. Which configuration step is needed to break the STP loop?
Configure a Layer 2 external bridged network on the interfaces facing the MST switches.
Enable the native VLAN on the interfaces facing the MST switches using static ports in a dedicated EPG.
Enable BPDU filter under the STP interface policy on the interfaces facing the MST switches.
Configure the STP instance to VLAN mapping under the switch STP policy.
Answer: B
The short version
B — Native VLAN on static ports in a dedicated EPG fixes the MST loop. ACI uses per-VLAN spanning tree while the outside MST region sends common spanning-tree BPDUs untagged on the native VLAN. Without native-VLAN provisioning those BPDUs are dropped, each MST switch claims root, and traffic loops through the fabric. Enabling native VLAN with static ports in a dedicated EPG restores BPDU exchange.
Key concepts in this question
MST BPDUs: Sent untagged on the native VLAN, unlike rapid per-VLAN frames, so trunks must carry untagged frames to peer.
Static-port EPG: Binds a VLAN to chosen leaf interfaces inside one endpoint group, giving external control frames a flood domain.
Dedicated EPG: Isolates MST-facing links from application groups so control flooding stays contained and predictable.
Loop cause: Missing BPDU visibility makes both switches forward, so restoring exchange lets blocking take effect.
Why B is correct
ACI must receive untagged MST BPDUs on the expected native VLAN to negotiate topology with the external region. Only a static-port assignment in a dedicated EPG provisions that native VLAN on the facing interfaces. Once forwarded between switches, the MST region sees one consistent neighbor, blocking engages, and the loop clears. That matches documented ACI to MST integration.
Why the others are wrong
A. external bridged network is a different integration pattern and does not provision native-VLAN BPDU handling on these ports.
C. BPDU filter suppresses processing entirely, hiding the loop while leaving a forwarding loop in place.
D. STP instance to VLAN mapping tunes internal ACI spanning tree without delivering untagged MST BPDUs.
300-620 exam tip
Link MST plus ACI to native VLAN on static ports, and link filter to danger.
2Which feature dynamically assigns or modifies the EPG association of virtual machines based on their attributes?
vzAny contracts
standard contracts
application EPGs
uSeg EPGs
Answer: D
The short version
D — uSeg EPGs bind VMs by attributes. Microsegmented EPGs classify endpoints dynamically from VM tags, names or OS facts instead of static port bindings. This matches the settled key and the vendor note on attribute based operation.
Key concepts in this question
uSeg EPG: a microsegmented group that selects endpoints with filters on VM attributes supplied by VMM integration.
VMM attribute source: vCenter or similar inventory feeds guest name, OS, tag and network facts into APIC for classification.
Standard EPG binding: regular application EPGs attach statically to ports, paths or VMM port groups without attribute matching.
Contracts and vzAny: policy objects that permit traffic between EPGs but do not assign membership, so they cannot replace classification.
Why D is correct
Only uSeg EPGs evaluate per-VM attributes at attach time and move the endpoint between groups when attributes change. Cisco ACI documentation describes uSeg as the attribute driven membership model tied to VMM domains. The question asks for dynamic assignment from attributes, which is exactly that behavior, so D is the decided key.
Why the others are wrong
A. vzAny simplifies contract scope across a VRF but never assigns EPG membership from VM facts.
B. Standard contracts enforce permit or deny between consumers and providers without classifying endpoints.
C. Application EPGs provide static grouping for ports and port groups, not dynamic attribute matching.
300-620 exam tip
When the stem says dynamically by attributes or VM tags, choose uSeg EPG and eliminate contracts and static EPGs immediately.
3An engineer must allow multiple external networks to communicate with internal ACI subnets. Which action should the engineer take to assign the prefix to the class ID of the external Endpoint Group?
Enable the Export Route Control Subnet for the External Endpoint Group flag.
Enable an L3Out with Shared Route Control Subnet.
Configure subnets with the External Subnets for External EPG flag enabled.
Configure subnets with the Import Route Control Subnet flag enabled.
Answer: C
The short version
C — External Subnets flag maps prefix to external EPG class. Enabling External Subnets for External EPG on the L3Out subnet tells ACI to program that prefix with the external EPG class ID. That linkage lets outside prefixes talk to internal ACI subnets under contract control.
Key concepts in this question
External EPG: a policy object representing outside prefixes reachable through an L3Out for contract enforcement.
External Subnets for External EPG flag: the subnet scope that binds a configured prefix to the EPG class ID in hardware.
Export and Import Route Control: separate scopes that govern route advertisement and import, not class ID assignment for data plane matching.
Shared Route Control: a scope for shared-services leaking between VRFs, not the general external to internal case.
Why C is correct
ACI programs external prefix to class mapping only when the subnet carries the External Subnets for External EPG scope. Without that flag the prefix may be advertised yet never matches the external EPG in policy lookup. The settled key therefore requires that exact flag to allow multiple external networks to reach internal subnets.
Why the others are wrong
A. Export Route Control governs prefix advertisement outward and does not assign the external class ID.
B. Shared Route Control enables shared L3Out patterns across VRFs, not standard external classification.
D. Import Route Control governs inbound route acceptance into the fabric, not EPG class binding.
300-620 exam tip
External prefix to class ID always means External Subnets for External EPG, while import or export alone mean routing control only.
4An engineer must ensure that Cisco ACI flushes the appropriate endpoints when a topology change notification message is received in an MST domain. Which three steps are required to accomplish this goal? (Choose three.)
Enable the BPDU interface controls under the spanning tree interface policy.
Configure a new STP interface policy.
Bind the spanning tree policy to the switch policy group.
Associate the STP interface policy to the appropriate interface policy group.
Create a new region policy under the spanning tree policy.
Map VLAN range to MST instance number.
Answer: C, E, F
The short version
CEF — MST flush needs region policy, VLAN mapping and bound STP policy. The three actions together let ACI track the MST region and clear only affected endpoints on topology change notifications. The community vote is unanimous for this triple and matches ACI MST interaction guides.
Key concepts in this question
MST region policy: defines name, revision and VLAN to instance mapping so ACI speaks the same region as external switches.
VLAN to MST instance map: tells ACI which VLANs belong to which instance so flush scope follows the correct spanning tree.
Switch policy group binding: applies the spanning tree policy to the right nodes so TCN handling is actually active.
Interface STP policy: tunes per-port BPDU behavior but alone cannot define region identity or instance mapping.
Why CEF is correct
ACI learns MST topology from region name plus revision plus instance mapping, then uses the bound switch policy to act on TCNs. Creating the region policy, mapping VLAN range to instance number, and binding the spanning tree policy to the switch policy group are each prerequisites for correct flushing. Omitting any one leaves region mismatch or inactive policy, so the triple CEF is required.
Why the others are wrong
A. BPDU interface controls tune guard and filter behavior but do not establish region or flush scope.
B. A new STP interface policy alone defines port settings without region identity or switch binding.
D. Associating an interface policy to an interface group affects ports only, not region-wide endpoint flush logic.
300-620 exam tip
MST in ACI is always region plus instance map plus bind to switch group, so look for the C E F triple on topology change questions.
5A Cisco ACI bridge domain and VRF are configured with a default data-plane learning configuration. Which two endpoint attributes are programmed in the leaf switch when receiving traffic? (Choose two.)
Remote MAC, IP
Remote Subnet
Local IP, not MAC
Local MAC, IP
Local Subnet
Remote IP
Answer: D, F
The short version
DF — default learning programs local MAC with IP plus remote IP. With default bridge domain and VRF data plane learning, the ingress leaf learns both MAC and IP for locally attached endpoints but only IP for routed remote sources. The settled DF pairing reflects switched versus routed learning rules.
Key concepts in this question
Local endpoint learning: a leaf that receives a frame on a front panel port can see source MAC directly and can learn IP when routing or ARP is involved.
Remote learning over VXLAN: a leaf receiving fabric traffic sees outer TEP addresses and inner payload, learning remote identity from inner headers.
Switched versus routed traffic: purely switched frames teach MAC only, while routed or ARP traffic also teaches IP binding.
Subnet based forwarding: remote subnet entries guide next hop selection but are not programmed as endpoint attributes in this default case.
Why DF is correct
Under default settings the leaf programs Local MAC plus IP for its own attached host because it sees both L2 source and, via SVI routing or ARP, the L3 binding. For traffic arriving from another leaf it programs Remote IP from the inner packet without creating a full remote MAC entry in this scenario. D plus F therefore describe exactly what the leaf writes on receipt.
Why the others are wrong
A. Remote MAC plus IP overstates what is learned for routed traffic in the stated default case.
B. Remote Subnet describes a route, not an endpoint attribute programmed on packet receipt.
C. Local IP without MAC is impossible for a directly attached host whose MAC is always visible.
E. Local Subnet is a bridge domain construct, not a per-endpoint learned attribute.
300-620 exam tip
Local means MAC plus IP together, remote in this default means IP alone, so DF beats any MAC-only or subnet choice.
6Refer to the exhibit. An engineer wants to initiate an ICMP ping from Server1 to Server2. The requirement is for the BD1 to enforce ICMP replies that follow the expected path. The packets must be prevented from taking the direct path from Leaf1 to Server1. Which action must be taken on BD1 to meet these requirements?
Set L2 Unknown Unicast to Flood.
Set L2 Unknown Unicast to Hardware Proxy.
Disable Unicast Routing.
Enable ARP Flooding.
Answer: C
The short version
C — Disable Unicast Routing on BD1 forces symmetric ICMP path. With routing off, BD1 bridges instead of routing between Server1 and Server2, so replies must return through the expected routed hop rather than shortcutting directly from Leaf1. That enforces the required path for inspection.
Key concepts in this question
Bridge domain unicast routing: when enabled the BD provides an anycast gateway and routes between subnets in hardware.
Symmetric path requirement: return traffic must follow the forward path so policy or service insertion sees both directions.
L2 Unknown Unicast handling: flood versus hardware proxy controls unknown destination flooding, not routed versus bridged forwarding.
ARP flooding: controls ARP distribution method and does not change whether inter-subnet traffic is routed.
Why C is correct
The exhibit shows two servers where direct leaf to server return would bypass the intended inspection point. Disabling unicast routing removes the local anycast route on BD1, forcing ICMP to stay at L2 and return via the designated gateway path. Only C changes forwarding from routing to bridging in the way the requirement demands.
Why the others are wrong
A. Flood for unknown unicast increases flooding without removing the routed shortcut for known IPs.
B. Hardware proxy optimizes unknown unicast lookup for anycast gateways but preserves routing behavior.
D. ARP flooding changes ARP propagation only and leaves unicast routing decisions untouched.
300-620 exam tip
To block a direct routed shortcut and force replies back through the gateway, disable unicast routing on that bridge domain.
7What is MP-BGP used for in Cisco ACI fabric?
MP-BGP VPNv4 AF is used as protocol on L3Out between a border leaf and an external router
MP-BGP Layer 2 VPN EVPN AF is used to propagate L3Out routes that are received from a border leaf
MP-BGP VPNv4 AF is used to propagate L3Out routes that are received from a border leaf to the fabric
MP-BGP VPNv4 AF is used between spines in an ACI Multi-Pod fabric to propagate the endpoint
Answer: C
The short version
C — VPNv4 distributes border L3Out routes fabric-wide. ACI uses MP-BGP inside the fabric to carry tenant routes learned at border leaves to every leaf that needs them. External prefixes arriving through an L3Out are imported into VPNv4 and reflected by spines to compute leaves. EVPN handles host reachability semantics, while L3Out IPv4 propagation is documented under VPNv4.
Key concepts in this question
L3Out: Defines external routing on border leaves, including protocols, interfaces, and external endpoint groups.
VPNv4: Carries IPv4 prefixes with distinguisher and target extensions, keeping tenant VRFs separate over one BGP core.
Route reflection: Spines reflect VPNv4 routes between leaves, avoiding full mesh while spreading border prefixes.
EVPN scope: Used for Layer 2 and host routes, not the documented carrier for L3Out external prefixes.
Why C is correct
Prefixes arrive at the border leaf via the L3Out peering and must reach destination leaves for hardware programming. Exporting them into VPNv4 with spine reflection achieves exactly that, matching the cited fabric BGP guide. No other family is documented as the L3Out to fabric carrier for these routes, so the settled statement follows ACI architecture directly.
Why the others are wrong
A. places VPNv4 on the external peering, but border to router uses the exterior protocol while VPNv4 stays internal.
B. assigns propagation to EVPN, misattributing the documented VPNv4 role for external prefixes.
D. limits VPNv4 to Multi-Pod endpoint sync, confusing inter-pod control with general L3Out distribution.
300-620 exam tip
Memorize L3Out learned at border plus VPNv4 equals fabric-wide external routes.
8An engineer must configure a group of servers with a contract that uses TCP port 80. The EPG that contains the web servers requires an external Layer 3 cloud to initiate communication. Which action must be taken to meet these requirements?
Configure the EPG as a provider and L3 out as consumer of the contract.
Configure OSPF to exchange routes between the L3 out and EPG.
Configure a taboo contract and apply it to the EPG.
Configure the EPG as a consumer and L3 out as a provider of the contract.
Answer: D
The short version
D — web EPG consumes, L3Out provides TCP 80. When the outside cloud initiates toward web servers, the direction of the contract must reflect initiator as consumer and server side as provider for that filter. The settled key reverses the intuitive provider choice to match ACI contract direction rules.
Key concepts in this question
Contract direction: provider exposes a service, consumer initiates toward it, filters define ports such as TCP 80.
EPG roles: web server EPG holds the servers, L3Out EPG represents the external cloud networks.
External initiation pattern: because the cloud starts the session, the L3Out must be the provider side for that flow.
Taboo and routing options: taboo denies traffic and OSPF exchanges routes, neither sets contract direction for TCP 80.
Why D is correct
ACI evaluates who provides the TCP 80 service as seen by the initiator. Here the external cloud needs to reach the web EPG, so the web EPG consumes the contract while the L3Out provides it. Configuring D programs the correct consumer provider pair and applies the TCP 80 filter in the right direction, which vendor discussion confirms after correcting the initial A guess.
Why the others are wrong
A. EPG as provider with L3Out as consumer models servers initiating outward, opposite to the stated external initiation.
B. OSPF exchanges routes for reachability but does not create the contract relationship for TCP 80.
C. Taboo contracts deny traffic and would block rather than permit the required web access.
300-620 exam tip
Initiator consumes, target provides, so when outside initiates inward the L3Out is provider and the server EPG is consumer.
9The unicast routing feature is enabled on the bridge domain. Which two conditions enable the Cisco ACI leaf to learn a source IP as a local endpoint? (Choose two.)
Through Ethernet traffic received in a bridge domain.
IP traffic routed through an SVI.
Through VXLAN traffic received on the uplink.
IP traffic routed through a Layer 3 Out.
Through ARP received on an SVI.
Answer: B, E
The short version
BE — SVI routed IP plus SVI ARP teach local IP. With unicast routing on the bridge domain, the leaf learns source IP as local only when it routes the packet through an SVI or snoops ARP on that SVI. Pure L2 switched frames teach MAC alone without IP binding.
Key concepts in this question
Unicast routing on BD: enables the pervasive gateway SVI on each leaf to route between subnets.
Routed IP learning: when the leaf performs the routing lookup, it can bind inner source IP to source MAC as local.
ARP on SVI learning: ARP requests handled by the SVI reveal IP to MAC mapping directly from ARP headers.
Switched and VXLAN cases: L2 only transit or fabric uplink receipt without SVI involvement never creates a local IP entry.
Why BE is correct
Cisco learning rules state the leaf learns MAC on any local receipt, but learns IP tied to MAC only from ARP headers or from IP headers of packets it routes. Both B and E involve the SVI in exactly those ways, so both create local IP entries. The vendor learning sequence quoted in discussion lists ARP plus routed IP as the two IP teaching paths.
Why the others are wrong
A. Ethernet switched traffic in a BD without ARP or routing teaches MAC only, never IP.
C. VXLAN on the uplink indicates remote arrival, which programs remote entries rather than local IP.
D. L3Out routed traffic arrives from outside and is learned as remote, not as a local endpoint.
300-620 exam tip
Local IP needs SVI involvement, so pick routed through SVI plus ARP on SVI and reject pure switched or uplink choices.
10When does the Cisco ACI leaf learn a source IP or MAC as a remote endpoint?
When VXLAN traffic arrives on a leaf fabric port from the spine and outer source IP is in the Layer 3 Out EPG subnet range.
When VXLAN traffic arrives on a leaf fabric port from the spine and outer source IP is in the bridge domain subnets range.
When VXLAN traffic arrives on a leaf fabric port from the spine and inner source IP is in the Layer 3 Out EPG subnet range.
When VXLAN traffic arrives on a leaf fabric port from the spine and inner source IP is in the bridge domain subnets range.
Answer: D
The short version
D — inner source IP in BD subnet means remote endpoint. When VXLAN arrives on a fabric port, the leaf looks past the outer TEP header to the inner source IP and learns it as remote if it belongs to a local bridge domain subnet. Outer TEP and L3Out ranges never define remote endpoint identity.
Key concepts in this question
VXLAN encapsulation: outer header carries spine and leaf TEP IPs for transport, inner header carries the true tenant source.
Inner versus outer IP: endpoint identity comes from inner source, while outer source only identifies the sending VTEP.
Bridge domain subnet scope: prefixes owned by the BD indicate tenant endpoints that may roam between leaves.
L3Out EPG subnet scope: outside prefixes learned via border leaves, not fabric remote endpoints in this context.
Why D is correct
The leaf must ignore outer TEP addressing and evaluate inner source against known BD subnets to decide remote learning. If inner IP falls inside the BD subnet range, the source lives behind another leaf and is programmed as remote. That inner plus BD subnet combination is exactly option D, confirmed by strong community consensus.
Why the others are wrong
A. Outer source in L3Out range describes transport from an external path, not tenant endpoint identity.
B. Outer source in BD range confuses TEP transport with tenant addressing and never occurs for valid fabric traffic.
C. Inner source in L3Out range indicates outside learned via border, not a fabric remote endpoint.
300-620 exam tip
For remote learning always read inner IP for identity and BD subnet for scope, and ignore outer TEP completely.