Enterprise Routing and Switching, Professional (JNCIP-ENT) — Free Practice Questions
10 free sample questions from a bank of 10, with the correct answers and explanations. No signup required — start practising right now.
1Click the Exhibit button.
You want to limit port access to only one device at a time.
Referring to the exhibit, which configuration change will accomplish this task?
Enable MAC RADIUS restrict.
Change the supplicant mode to multiple.
Change the supplicant mode to single-secure.
Change the maximum EAPOL request to 1.
Answer: C
The short version
C — Change the supplicant mode to single-secure.
802.1X supplicant mode decides how many devices may be attached behind an authenticator port; single-secure admits exactly one authenticated device per port, which is what "only one device at a time" requires.
Key concepts
802.1X supplicant mode: the Junos dot1x authenticator setting on an access interface that governs how many hosts may use the port after authentication (multiple, single, single-secure, mac-limit).
single-secure: only one device is allowed to authenticate on the port; the port is secured for that one supplicant, and re-authentication of a different device is rejected until the current one goes away.
multiple: any number of devices may authenticate on the port — this is the opposite of the requirement.
MAC RADIUS: a separate Layer 2 authentication method that authenticates by MAC address via RADIUS; mac-radius can be combined with 802.1X but is not a port-occupancy limit.
Retransmission count: the number of EAPOL/identity retries the authenticator sends while waiting for a supplicant — a liveness timer, not an access policy.
Why C is correct
The behaviour you need to constrain is how many supplicants may be active on the port. That is precisely the parameter the supplicant mode controls. Setting supplicant-mode single-secure tells the authenticator to allow exactly one authenticated device on the interface: the first supplicant that completes 802.1X is the only one permitted, and a second device on the same port cannot authenticate while it is attached. That is the direct, purpose-built mechanism for "only one device at a time", and it is why the answer is the supplicant mode and not any of the RADIUS or retry settings.
Why the others are wrong
A. MAC RADIUS authenticates a device by its MAC address, but it does not limit the port to a single device — MAC RADIUS on its own authenticates every learned MAC it is presented with. It changes how a device is identified, not how many devices may use the port, so it cannot accomplish the task.
B.multiple is the exact opposite of what is required: it lets any number of devices authenticate on the port, which would permit precisely the multi-device access you are trying to prevent.
D. The maximum EAPOL retransmission count only controls how many times the authenticator retries EAPOL-Identity/Start while waiting for the supplicant to respond. It affects how long the switch waits before declaring the port dead; it has no bearing on how many devices may be admitted.
JN0-650 exam tip
Port occupancy = supplicant mode (multiple / single / single-secure / mac-limit). Retry counts and MAC RADIUS affect identity and timing, never how many devices a port admits.
2Your OSPF network consists of a mix of 1GbE and 10GbE interfaces. By default, all interfaces have the same cost in your OSPF network. You are asked to ensure that the 10GbE interfaces are more preferred when available.
In this scenario, which two statements would accomplish this behavior? (Choose two.)
You should define the reference bandwidth as 10G, which will assign the 1GbE interfaces a higher cost.
You should manually assign the interface metric for each 10GbE interface to be higher than the 1GbE interfaces in your OSPF network.
You should define the reference bandwidth as 1G, which will assign the 1GbE interfaces a higher cost.
You should manually assign the interface metric for each 1GbE interface to be higher than the 10GbE interfaces in your OSPF network.
Answer: A, D
The short version
A and D — Set the OSPF reference bandwidth to 10G, or raise the interface cost on the 1GbE interfaces.
OSPF metric is computed as reference bandwidth divided by interface bandwidth, so with a 10G reference the 1G links get a cost of 10 and the 10G links a cost of 1 — or you can achieve the same skew manually by giving the 1G interfaces a higher metric.
Key concepts
OSPF cost: the interface metric derived as reference bandwidth / interface bandwidth, added to reach the router ID at each hop.
reference bandwidth: the base bandwidth assumed to be the fastest link in the network; set with set protocols ospf reference-bandwidth, and the highest-speed interface bandwidth by default.
Higher cost = less preferred: OSPF's shortest-path first algorithm sums interface costs, so expensive links are avoided as long as an alternative exists.
Manual interface metric: set protocols ospf area X interface Y metric Z overrides the computed cost for that interface.
Junos default: without an explicit reference bandwidth, all interfaces whose bandwidth is at or below the default fastest link compute to cost 1, which is why 1G and 10G look identical by default.
Why A and D are correct
A. OSPF derives cost as reference bandwidth divided by interface bandwidth. Raising the reference to 10G gives a 1G interface a cost of 10G/1G = 10 and a 10G interface a cost of 1, so the SPF automatically prefers the 10G path whenever one exists while still using 1G for backup. This is the scalable, self-maintaining way to express the requirement.
D. Assigning each 1GbE interface a metric greater than the 10G interfaces produces exactly the same preference in the SPF calculation. It is the fallback when you cannot or do not want to change the global reference bandwidth, at the cost of configuring every interface individually and keeping it in sync.
Why the others are wrong
B. Giving the 10GbE interfaces a higher metric makes the fast links the expensive ones, which is the exact inverse of the requirement — the SPF would prefer 1GbE whenever a path exists.
C. Defining the reference bandwidth as 1G assigns a cost of 1 to the 1GbE interfaces and a lower-than-1 cost (0, which OSPF treats as the minimum) to the 10G links. It is true this makes 1G look expensive, but the reasoning and framing in the option claim 1G gets a higher cost, whereas the meaningful configuration change for this ask is to reference the fastest link and let the slower links be penalized — with a 1G reference the 10G links all collapse to the same minimum cost and there is no relative metric left to model them.
JN0-650 exam tip
OSPF cost = reference bandwidth / interface bandwidth, and a higher cost means less preferred. To favour fast links: raise the reference bandwidth or raise the slow interfaces' metrics — never the other way round.
3You have two multicast receivers connected to the same VLAN. You notice that the switch that they are connected to is forwarding multicast traffic out of all the ports in the same VLAN, instead of just the two ports for the connected multicast receivers.
In this scenario, what would you configure to optimize multicast forwarding?
promiscuous mode
spanning tree
IGMP snooping
IRB interface
Answer: C
The short version
C — Enable IGMP snooping.
IGMP snooping lets the switch inspect the multicast membership reports and build a forwarding table so multicast traffic leaves only through ports that actually have receivers, instead of flooding the VLAN.
Key concepts
IGMP snooping: a Layer 2 function where the switch listens to IGMP membership reports and leaves between hosts and the multicast router, and uses them to build a per-VLAN multicast forwarding table.
Multicast flooding: the default behaviour, where a multicast frame is replicated out every port in the VLAN because the switch has no idea where the receivers are.
Querier: the router (or a switch configured as a querier) that sends IGMP queries and maintains group state; with snooping, a querier must be present for reports to be answered.
mrouter port: a port leading toward the multicast router, which snooping learns so upstream traffic is not pruned away.
IGMPv3 / SSM: report types that also carry source-specific group information, giving snooping finer forwarding granularity.
Why C is correct
By default a switch has no idea which hosts have joined a multicast group, so it must treat multicast like broadcast and flood it out every port in the VLAN. IGMP snooping removes that guesswork: the switch listens in the middle of the IGMP exchange between hosts and the multicast router, records which ports sent membership reports for which groups, and builds a multicast forwarding table. With that table, a multicast frame is replicated only to the ports that have receivers — here, the two ports holding the receivers — and not to the rest of the VLAN. That is exactly the inefficiency described in the question and exactly what IGMP snooping fixes.
Why the others are wrong
A. Promiscuous mode is a host-side setting (or a mirroring-related option) that lets a device see all frames arriving on an interface. It has no effect on how the switch forwards multicast traffic and cannot build receiver-aware forwarding state.
B. The spanning tree controls loop avoidance and topology by forwarding unicast/broadcast per the tree. Multicast flooding as described is a VLAN-wide replication issue, not a loop issue, so the spanning tree is unrelated.
D. An IRB interface is the integrated routing and bridging interface that provides the Layer 3 gateway for a VLAN. It is where multicast routing happens at Layer 3, but it does nothing about a Layer 2 switch flooding a group out of every port of the same VLAN.
JN0-650 exam tip
Flooding a group to every port in the VLAN = missing IGMP snooping. Snooping needs a querier; the fix is at Layer 2, not in the IRB or the spanning tree.
4You are implementing an EVPN-VXLAN edge-routed bridging design using Layer 3 gateway operations.
In this scenario, which statement is correct?
Each distribution switch has unique IP addresses for IRB interfaces; routing protocols run on IRB interfaces.
Distribution switches share the same anycast IP addresses for IRB interfaces; routing occurs at the distribution layer.
Only core switches have IRB interfaces; all Layer 3 routing happens in the core.
IRB interfaces are disabled; all routing happens through external routers only.
Answer: B
The short version
B — The distribution switches share the same anycast IP addresses for their IRB interfaces, and routing happens at the distribution layer.
This is the anycast-gateway model: a single Layer 3 gateway address (with its anycast MAC) is configured on every leaf, so traffic is locally routed no matter which leaf a server lands on.
Key concepts
EVPN-VXLAN: a VXLAN overlay carried over BGP EVPN, providing both Layer 2 (Ethernet segment) and Layer 3 (IP prefix) services in a fabric.
IRB interface: the integrated routing and bridging interface that provides the gateway IP and MAC for a VLAN, terminating routing locally.
Anycast gateway: the same gateway IP/MAC configured on all leaf IRB interfaces, so the default gateway is reachable from any leaf without relying on the underlay.
Routing at the distribution layer: fabric-wide routing (EVPN IP prefixes, typically advertised as Type 5 or type-2 route targets) learned via the spines and distributed to the leaves.
Layer 3 gateway operations: the design choice that moves the gateway responsibility into the fabric, with the distribution layer running the routing protocol for the underlay and advertising reachability.
Why B is correct
With Layer 3 gateway operations in an EVPN-VXLAN fabric, the leaves do not need a unique gateway per leaf. Instead the same anycast gateway IP address (and its anycast MAC) is configured on every leaf's IRB interface. A server can then move to any leaf and its default gateway remains valid, and ingress traffic from an external source is load-balanced across leaves because every leaf answers for that address. Layer 3 routing for the fabric — underlay prefixes, EVPN IP routes, and the default route out — is terminated and distributed at the distribution layer, which is what advertises the prefixes that the leaf IRBs install. Sharing the anycast address is therefore the defining characteristic of this design, not per-leaf unique addressing.
Why the others are wrong
A. Unique IRB IP addresses per distribution switch describe a design where each layer-3 gateway is a distinct host, i.e. routing not being shared. That is the opposite of the anycast-gateway model used with Layer 3 gateway operations, and it loses the ability for a server to roam between leaves.
C. Restricting IRB interfaces to the core switches and doing all routing in the core is a traditional routed design that contradicts having leaf-level Layer 3 gateway operations in the fabric; the leaves would have to hairpin every frame to the core.
D. Disabling IRB interfaces and relying on external routers removes the distributed gateway from the fabric entirely, which is the opposite of "implementing EVPN-VXLAN edge-routed bridging using Layer 3 gateway operations".
JN0-650 exam tip
Anycast gateway = one gateway IP/MAC on every leaf IRB. If you hear "same IP on all distribution switches", the answer is anycast gateway with routing at the distribution layer.
5Which statement about LLDP and LLDP-MED operations on EX Series devices is correct?
LLDP only operates on interfaces configured for Layer 2.
EX Series devices flood LLDP frames across a Layer 2 domain to calculate a network topology.
EX Series devices support LLDP-MED power negotiation, enabling dynamic allocation of PoE power based on endpoint device needs.
LLDP-MED focuses on discovering network connectivity devices like routers and switches.
Answer: C
The short version
C — LLDP-MED power negotiation lets EX Series switches allocate PoE power dynamically based on the endpoint's actual need.
LLDP-MED extends LLDP with media/end-point specific TLVs including power-over-Ethernet information, so a phone or AP can advertise its class and the switch returns only the power it requires.
Key concepts
LLDP (Link Layer Discovery Protocol): IEEE 802.1AB, a link-local (destination MAC 01:80:C2:00:00:0E) protocol by which devices advertise identity, capabilities and port information to directly connected neighbours.
LLDP-MED: the Media Endpoint Discovery extension (TIA-1059) that adds TLVs for network connectivity device policy/location and for endpoint media capabilities.
LLDP-MED power / extended PoE: the Power Via MDI TLV by which an endpoint advertises its power source, power priority and power requirements, allowing the switch to grant exactly what is needed.
PoE classes: legacy class 0-3 classification by signature/resistance, and the LLDP-MED class 1-8 classification by data type, which makes dynamic allocation possible.
LLDP-MED network connectivity device: the optional LLDP-MED device type that advertises VLAN/port voice policy so endpoints can select the right VLAN.
Why C is correct
Power over Ethernet on an EX Series switch can be allocated statically (fixed wattage per port) or dynamically. With LLDP-MED enabled, an endpoint such as an IP phone, wireless AP or access point advertises its power source, priority and required power in LLDP-MED TLVs as part of its normal LLDP announcements. The switch reads that advertisement, allocates the matching amount of power from the port's PoE budget, and reserves the remainder for other ports. Because this exchange happens continuously as endpoints attach and detach, PoE power is dynamically allocated to actual demand rather than statically over-provisioned port by port — which is exactly what the correct statement describes.
Why the others are wrong
A. LLDP is not restricted to Layer 2 ports. It runs on Layer 2 and Layer 3 routed interfaces alike, as long as the interface is up and link-local adjacency is possible — the protocol is independent of the forwarding mode.
B. LLDP frames are sent to a link-local multicast address and are not flooded across the Layer 2 domain to build a topology. LLDP is neighbour-to-neighbour on each port; no switch computes a topology by flooding it.
D. LLDP-MED's original focus is endpoint/media devices — phones, wireless access points and other customer premises equipment — and it extends LLDP for those devices. Discovery of network connectivity devices (routers, switches) is plain LLDP; LLDP-MED may optionally add policy/location information for them, but that is not its defining purpose.
JN0-650 exam tip
LLDP = neighbour identity, link-local only. LLDP-MED = endpoint-focused extension, and its most exam-worthy use is dynamic PoE allocation via the power TLVs.
6Click the Exhibit button.
Referring to the exhibit, leaf2 is elected the designated forwarder.
Which two statements are correct if server2 generates a broadcast frame? (Choose two.)
The broadcast frame will be forwarded to leaf1 and leaf2.
The broadcast frame will not be forwarded to leaf1.
Only leaf2 will forward the broadcast frame to server1.
Only leaf1 will forward the broadcast frame to server1.
Answer: A, C
The short version
A and C — The broadcast is flooded to both leaf1 and leaf2, but only leaf2 forwards it toward server1.
Leaf2 is the designated forwarder for the segment, so it originates the copy into the overlay; leaf1 receives it as a BUM frame but must not flood it back into the same segment.
Key concepts
BUM traffic: broadcast, unknown-unicast and multicast frames, which must be flooded across the broadcast domain in any Layer 2 fabric.
All-active (anycast) VXLAN: the design where every leaf originates BUM traffic for a segment, so there is no single forwarding bottleneck.
Designated forwarder (DF): in all-active mode the role is elected and shared; the DF is responsible for originating the BUM frame for the segment, typically using a designated-forwarder election based on node IP and tunnel key.
BUM policing / split horizon: leaf switches suppress re-flooding of a BUM frame they receive from the overlay for the same segment, which prevents a forwarding loop.
Source-based learning / hairpin: the remote copy arriving at leaf1 is a learned unicast from leaf2's tunnel endpoint, so leaf1 does not retransmit it.
Why A and C are correct
A. A broadcast frame from server2 must be delivered to every other host in the broadcast domain, so it must cross the VXLAN tunnel to all other leaves — including leaf1. Both leaves therefore receive a copy; anything less would break broadcast semantics.
C. Leaf2 is elected designated forwarder for the segment, so it is the leaf that originates the BUM copy into the overlay. The copy arriving at leaf1 is treated as a unicast frame from leaf2's tunnel endpoint and, because it already belongs to leaf1's local segment, leaf1 does not flood it again — so only leaf2 puts the broadcast onto the local access segment toward server1.
Why the others are wrong
B. Incorrect because the broadcast does reach leaf1. VXLAN BUM flooding crosses the tunnel to every leaf in the broadcast domain; suppressing delivery to a leaf would mean server1 on leaf1 could never receive broadcasts.
D. Reverses the forwarding roles. Leaf1 is not the designated forwarder here — leaf2 is — and it is precisely leaf1's behaviour to not re-flood a BUM frame back into the segment it came from.
JN0-650 exam tip
In an all-active VXLAN segment, the designated forwarder originates BUM traffic and the receiving leaf does not re-flood it. Flood to everyone; originate from the DF only.
7Which two statements correctly describe how EX Series switches use captive portal for Layer 2 authentication? (Choose two.)
The captive portal is the default Layer 2 authentication method that is applied before other methods such as 802.1X or MAC RADIUS.
The captive portal authentication allowlist works for devices that do not have HTTP capabilities.
The captive portal is configured on Layer 3 interfaces and does not participate in Layer 2 authentication on EX Series switches.
The captive portal is used as a fallback mechanism for clients that fail 802.1X or MAC RADIUS authentication.
Answer: B, D
The short version
B and D — The captive portal allowlist covers non-HTTP devices, and the captive portal acts as a fallback for clients that fail 802.1X or MAC RADIUS.
EX Series runs the portal as one Layer 2 authentication method in the access-profile order, and the allowlist lets pre-provisioned or non-browser endpoints through the portal step.
Key concepts
Captive portal (Layer 2): on EX Series, an authentication method in the access profile in which the switch redirects the client's HTTP request to a locally hosted or external portal for credentials before admitting it.
Access profile / authentication order: the ordered list of authentication methods applied to an interface, for example [ captive-portal mac-radius 802.1x ], evaluated until one succeeds or all fail.
Allow list: a set of MAC addresses (or filters) exempt from the captive portal redirect, intended for devices that cannot run an HTTP client — phones, printers, IoT endpoints.
Fallback behaviour: when an earlier method in the order rejects or cannot handle a client, evaluation proceeds to the next one; a device with no browser fails the portal and is offered the remaining methods.
802.1X / MAC RADIUS: EAP-based and MAC-based authentication methods that do not require the endpoint to speak HTTP, which is why the allowlist and fallback role belong to the portal.
Why B and D are correct
B. The allow list for non-HTTP devices: The captive portal works by intercepting the client's initial HTTP request and returning a redirect to a login page. A device with no HTTP client — a VoIP phone, a printer, a camera — can never complete that exchange, so it would fail the portal step even with correct credentials. The allowlist exempts exactly those devices from the portal redirect so they can fall through to a method that suits them.
D. The fallback after 802.1X/MAC RADIUS: Because the portal is one method in the interface's authentication order, it is commonly placed as the last resort. A client that fails 802.1X or MAC RADIUS then hits the portal and is redirected to a login page, giving it a manual authentication path instead of being dropped with no recourse.
Why the others are wrong
A. The captive portal is not the default method applied before everything else. Its position in the access profile is a design decision, and putting the portal first would break every endpoint that cannot complete an HTTP redirect.
C. This inverts where the captive portal lives. On EX Series it is a Layer 2 authentication method configured on the access interface inside the access profile, which is exactly why the allowlist and fallback behaviours in B and D exist at all.
JN0-650 exam tip
Captive portal = HTTP redirect, so it needs a browser. It earns its keep at the end of the authentication order as a fallback, and its allowlist rescues devices with no HTTP client.
8Click the Exhibit button.
Referring to the exhibit, which two statements are correct? (Choose two.)
R1 will advertise the 203.0.113.0/24 route as an OSPF Type 7 LSA into Area 0.
R5 will advertise the 203.0.113.0/24 route as an OSPF Type 5 LSA into Area 1.
R1 will advertise the 203.0.113.0/24 route as an OSPF Type 5 LSA into Area 0.
R5 will advertise the 203.0.113.0/24 route as an OSPF Type 7 LSA into Area 1.
Answer: C, D
The short version
C and D — R1 originates 203.0.113.0/24 as a Type 5 external LSA into Area 0, and R5 originates it as a Type 7 NSSA-external LSA into Area 1.
203.0.113.0/24 is redistributed into OSPF as an external route, and each area receives it in the LSA format its own area type permits: normal area gets Type 5, NSSA gets Type 7.
Key concepts
External route: a route learned from outside OSPF or redistributed from another protocol into OSPF; it is flooded as an AS external LSA.
Type 5 (AS External): flooded by an ASBR throughout all normal areas, but never into an NSSA or a stub area.
Type 7 (NSSA External): originated inside an NSSA by a router that has the route; translated into a Type 5 by an ABR when it crosses into a normal area.
NSSA: a not-so-stubby area, where external routes are allowed but the Type 5 is blocked from entering the area, so externals appear only as Type 7.
ABR / ASBR: an Area Border Router connects areas and performs LSA type translation; an Autonomous System Boundary Router originates external LSAs.
Why C and D are correct
C. R1 originates it as a Type 5 into Area 0: 203.0.113.0/24 is redistributed into OSPF, so R1 is the ASBR for that prefix and originates it as an AS external LSA, Type 5. Type 5 LSAs are flooded by any ASBR into every normal (non-stub, non-NSSA) area, so inside Area 0 the route arrives as a Type 5.
D. R5 originates it as a Type 7 into Area 1: Area 1 is an NSSA. Type 5 LSAs are deliberately kept out of an NSSA, so inside Area 1 the only way to carry an external route is as a Type 7 (NSSA external) LSA. R5, having learned the redistributed prefix, originates the Type 7 into Area 1, and it is translated into a Type 5 only when an ABR propagates it out of the NSSA into a normal area.
Why the others are wrong
A. A Type 7 LSA is never flooded into a normal area. It exists only inside an NSSA and must be translated into a Type 5 by an ABR before it can leave, so R1 does not originate a Type 7 into Area 0 — that area gets the external route as a Type 5.
B. A Type 5 LSA is not advertised into an NSSA; that restriction is the entire reason NSSAs exist. Inside Area 1 the external route must be carried as a Type 7 originated by R5, never as a Type 5 flooded from elsewhere.
JN0-650 exam tip
Normal area gets externals as Type 5; NSSA gets them as Type 7 and an ABR translates 7 to 5 on the way out. Stub areas get neither.
9You are deploying a new campus switching environment using EX Series switches. This new environment includes IP phones requiring Power over Ethernet (PoE) and end-user client PCs attached to a single switch port that will be configured with 802.1X authentication.
Considering the default configuration on an EX Series access switch and the configuration requirements to support this deployment, which two statements are correct? (Choose two.)
Configuration for the voice VLAN for all relevant access ports must be added.
Configuration for PoE on all relevant access ports must be added.
Configuration for LLDP and LLDP-MED on all relevant access ports must be added.
Configuration for 802.1X on all relevant access ports must be added.
Answer: A, D
The short version
A and D — You must add the voice VLAN configuration and the 802.1X configuration on the relevant access ports.
PoE and LLDP/LLDP-MED are already enabled by default on EX Series access interfaces, so only the voice VLAN and the authentication method need explicit configuration.
Key concepts
Voice VLAN: the separate VLAN assigned to an IP phone port so phone traffic is carried in its own VLAN; configured per access interface (for example voice-vlan), often with a voice VLAN ID and optional LLDP-MED voice policy.
802.1X: the port-based authentication method that must be configured on the interface with an access profile before supplicants on that port are challenged.
PoE: Power over Ethernet is enabled by default on EX Series access ports, so powering the phones requires no additional configuration on a modern EX access switch.
LLDP / LLDP-MED: also enabled by default on EX access interfaces, which is how phones discover the voice VLAN and negotiate power without extra configuration.
Access profile: the authentication container (for example [ mac-radius 802.1x ]) referenced by the interface to define which Layer 2 authentication methods apply.
Why A and D are correct
A. An IP phone and a PC share one access port. To put the phone's traffic into the voice VLAN while the PC stays in the data VLAN, the interface must be told about the voice VLAN. That is additional configuration that does not exist by default, so it must be added for this deployment to work as described.
D. The requirement states the ports will use 802.1X authentication, and no authentication method is enabled on an EX access port by default. The interface must therefore be configured for 802.1X with an appropriate access profile before the clients are challenged.
Why the others are wrong
B. PoE is enabled by default on EX Series access interfaces. Powering IP phones on those ports requires no added configuration, so listing it as a required change is incorrect — unless the ports had been explicitly disabled, which the question does not state.
C. LLDP and LLDP-MED are also enabled by default on EX access interfaces; they are how the phone discovers the voice VLAN and negotiates power. No additional configuration is needed for them to operate.
JN0-650 exam tip
On an EX access switch the defaults are PoE-on and LLDP-on; what is off by default is the voice VLAN and 802.1X. Only those two need configuration.
10Which two statements are correct about multicast routing tables? (Choose two.)
inet.2 is used when the multi cast forwarding topology differs from unicast forwarding.
MBGP and IS-IS place routes in inet.1 directly.
inet.0 is used for RPF checks.
IS-IS and OSPFv3 place routes in inet.2 directly.
Answer: A, C
The short version
A and C — inet.2 is used when the multicast forwarding topology differs from unicast, and inet.0 supplies the RPF check.
Multicast can need routes that unicast does not; those go in the alternate multicast table inet.2, while the RPF lookup that validates the source of a multicast packet is done against the unicast table.
Key concepts
RPF (Reverse Path Forwarding): the check that the source address of a multicast datagram arrived on an interface that would be used to reach that source; if not, the packet is discarded.
inet.0: the primary IPv4 unicast routing table; it holds the routes used both for unicast forwarding and as the RPF lookup source for multicast.
inet.2: the alternate multicast routing table, populated when the multicast forwarding topology must be different from the unicast topology.
inet.1: the primary multicast routing table used when multicast routing is kept in a table of its own rather than alongside unicast.
Multicast routing protocols: protocols such as PIM, MSDP and MBGP install multicast routes; on Junos their routes may go into inet.1/inet.2 by configuration.
Why A and C are correct
A. By default multicast RIBs are kept separate from unicast. When the design requires the multicast forwarding path to differ from unicast — different sources, a different set of outgoing interfaces, asymmetric routing — the multicast routes are installed in the alternate multicast table inet.2 rather than being mixed into inet.0.
C. RPF looks up the source address of a multicast packet in the unicast routing table to confirm the packet arrived on the correct interface. On Junos that RPF lookup is performed against inet.0, the primary unicast table, because the unicast topology is the reference for the check.
Why the others are wrong
B. MBGP and IS-IS do not install their multicast routes directly into inet.1 in this way. IS-IS is a unicast link-state IGP used as the underlay (and PIM uses it for RPF), and MBGP multicast routes are handled through the multicast table configured for the design — inet.1 or inet.2 — not by IS-IS placing anything there.
D. OSPFv3 and IS-IS are unicast IGPs. They populate unicast tables and provide the RPF information for multicast routing; they do not place multicast routes directly into inet.2. inet.2 is populated by multicast routing mechanisms configured to use it.
JN0-650 exam tip
RPF looks in inet.0 (unicast reference); multicast routes for a divergent topology go in inet.2. IGPs feed multicast forwarding, they do not own inet.2.