JN0-650 Exam Dumps - Enterprise Routing and Switching - Professional (JNCIP-ENT)

The IETF standards for EVPN-VXLAN define the scalability and forwarding behavior of the architecture.

VXLAN Scale (Statement B): The Virtual Network Identifier (VNI) is a 24-bit field in the VXLAN header. This allows for up to $2^{24}$ unique identifiers, which translates to approximately 16 million distinct VXLAN broadcast domains (segments) within a single fabric. * Forwarding Behavior (Statement D): VXLAN is a MAC-in-UDP encapsulation technology. When a Layer 2 frame enters the tunnel at a VTEP, the original Layer 2 frame (including the original source and destination MAC addresses) is encapsulated into a UDP/IP packet and transported across the IP underlay. When it reaches the egress VTEP, the header is removed and the Layer 2 frame is forwarded unchanged to the destination host.

Why others are incorrect: Statement A is incorrect because EVPN itself can technically support more than 16 million domains, but the data plane (VXLAN) is the limiting factor in this specific deployment scenario. Statement C is incorrect because VXLAN encapsulates Layer 2 frames, not just " IP packets, " and it treats the entire frame as a payload.

The exhibit illustrates an EVPN-VXLAN fabric using multi-homing. Server1 is dual-homed to Leaf1 and Leaf2 using an Ethernet Segment Identifier (ESI) in an active-active or active-standby configuration. Server2 is single-homed to Leaf3.

In EVPN environments, specific mechanisms are used to prevent loops and duplicate traffic when dealing with Broadcast, Unknown Unicast, and Multicast (BUM) traffic:

Broadcast Propagation (Option A): When Server2 (connected to Leaf3) generates a broadcast frame, Leaf3 encapsulates this traffic into a VXLAN tunnel and forwards it to all other VTEPs (Virtual Tunnel End Points) that participate in that specific Virtual Network Instance (VNI). Therefore, both Leaf1 and Leaf2 will receive the broadcast frame from the core fabric.

Designated Forwarder (DF) Election (Option C): To prevent a multi-homed device (like Server1) from receiving duplicate copies of the same broadcast frame from multiple leaf switches, EVPN utilizes a Designated Forwarder (DF). For a given ESI, only one leaf switch is elected as the DF.

The exhibit explicitly states that Leaf2 is the Designated Forwarder.

According to the Junos EVPN implementation, only the DF is responsible for decapsulating BUM traffic received from the core and forwarding it to the local access segment.

While Leaf1 receives the frame (Option A), it will see that it is not the DF for that segment and will drop the frame rather than forwarding it to Server1. This ensures Server1 receives exactly one copy of the broadcast, delivered by Leaf2.

Option B is incorrect because in a standard ingress replication or multicast-based underlay, all VTEPs in the VNI receive the broadcast.

Option D is incorrect because Leaf1 is the non-designated forwarder (non-DF) for this segment and is prohibited from forwarding BUM traffic to the ESI.

In Junos OS Class of Service (CoS), schedulers are the fundamental components used to manage the resources of individual egress queues. They determine how the switch handles traffic for a specific forwarding class as it waits to exit an interface.

Buffer Size (Option A): A scheduler defines the buffer size (memory allocation) for its assigned queue. This buffer holds packets during periods of congestion to prevent immediate packet loss. The size can be configured as a specific percentage of the total interface buffer, a temporal value in microseconds, or using the " remainder " of available space.

Queue Priority (Option D): A scheduler identifies the priority of the queue. Junos supports multiple priority levels, such as low , medium-low , medium-high , and high . This setting dictates the order in which the transmission hardware services the queues; for example, high-priority queues are typically emptied before low-priority ones.

Why other options are incorrect: Interface rate-limiting (Option B) is typically managed at the interface level or through policers , not the scheduler itself, which focuses on queue-specific transmission rates. DiffServ code translation (Option C) is the responsibility of rewrite rules , which modify the packet header markings as they leave the switch.

Junos OS 24.4 on EX Series switches provides robust support for LLDP (Link Layer Discovery Protocol) and its extension, LLDP-MED (Media Endpoint Discovery).

LLDP-MED Power Negotiation: This feature allows a switch (Power Sourcing Equipment or PSE) and a connected device (Powered Device or PD), such as an IP phone or access point, to negotiate power requirements beyond the standard IEEE 802.3af/at classes. The switch can dynamically allocate the exact amount of power the device needs (in 0.1W increments), which optimizes the power budget of the switch.

LLDP Scope: LLDP is a Link Layer protocol (Layer 2), but it is not restricted to Layer 2 interfaces; it can also operate on Layer 3 interfaces to advertise system identity and capabilities. This makes Option A incorrect.

Link-Local Protocol: LLDP frames use a specific multicast MAC address (01:80:c2:00:00:0e) that is not flooded or forwarded by switches. They are strictly link-local between two directly connected neighbors. This makes Option B incorrect.

Endpoint Focus: LLDP-MED is specifically designed for Media Endpoint Devices (like VoIP phones), providing TLVs for network policy (VLAN/QoS), location identification, and inventory management. Standard LLDP is used for discovering network connectivity devices. This makes Option D incorrect.

In a three-stage EVPN-VXLAN fabric (Spine-Leaf architecture) where leaf devices perform the gateway services (Edge-Routed Bridging or ERB):

VTEP Functionality (Option B): Leaf devices act as VXLAN Tunnel End Points (VTEPs). They are responsible for encapsulating Ethernet frames from locally connected hosts into VXLAN packets for transport across the IP fabric and de-encapsulating incoming VXLAN packets.

Fabric Support (Option C): In a standard Juniper IP fabric design, both the leaf and spine devices are typically expected to support VXLAN and EVPN capabilities. While spines in an ERB design might only perform IP forwarding in the underlay, they often participate in the BGP control plane (as Route Reflectors) or provide the ability to transition to Centrally-Routed Bridging (CRB) where spines would act as gateways.

Incorrect Options: Option A is incorrect because the entire fabric infrastructure must be " VXLAN aware " in terms of MTU settings and routing capability to support the overlay. Option D is incorrect because in an ERB design, the spine devices are high-speed transit nodes and do not terminate VXLAN tunnels; only the edge (leaf) devices act as VTEPs.

Junos OS 24.4 handles multicast traffic using two distinct models based on whether the source is known in advance.

Knowledge and Sources (Option A): PIM-ASM (Any Source Multicast) is designed for " many-to-many " communication where receivers join a group ($*,G$) and rely on a Rendezvous Point (RP) to discover active sources. In contrast, PIM-SSM (Source-Specific Multicast) is for " one-to-many " scenarios where receivers must already know the exact source IP ($S,G$) before joining.

Protocol and RP Logic (Option D): PIM-SSM bypasses the RP entirely. It relies on IGMPv3 messages from the host, which explicitly include both the source address and the group address. This allows the last-hop router to build a Shortest Path Tree (SPT) directly toward the source immediately.

Register Process (Option B): While PIM-ASM does use registers, it is the First-Hop Router (Designated Router on the source segment) that sends the register to the RP, not the receiver ' s DR.

RP Role (Option C): This is exactly backwards; PIM-ASM requires an RP for source discovery, whereas PIM-SSM does not use one at all.

In an EVPN-VXLAN Edge-Routed Bridging (ERB) architecture, also known as a collapsed fabric, the Layer 3 default gateway functionality is moved from the core/spine layer down to the edge (the leaf or distribution layer).

Anycast Gateways: To support seamless host mobility and redundancy, multiple distribution/leaf switches are configured with the same anycast IP address and MAC address on their IRB interfaces for a given VLAN. This allows a host to move between different switches without needing to update its default gateway configuration or ARP cache.

Distributed Routing: Routing occurs locally at the edge (distribution layer). Traffic destined for a different subnet is routed by the first switch it hits (the ingress leaf), rather than being backhauled to a central core router.

Symmetric vs. Asymmetric IRB: Junos OS 24.4 supports both models, but the ERB design typically utilizes symmetric routing for better scalability, where each leaf only needs to know the routes for its locally connected VNIs and uses a transit VNI for inter-subnet communication.

Option A is incorrect because while unique IPs can be used (Method 1 in some docs), the defining characteristic of an efficient ERB design is the use of shared Anycast IPs for the gateway. Option C describes a Centrally-Routed Bridging (CRB) design, not ERB.

VSTP (VLAN Spanning Tree Protocol) is Juniper ' s implementation that provides a separate spanning tree instance for each VLAN, ensuring compatibility with Cisco ' s PVST+. However, it has significant scaling limitations:

Instance Limits: On many Juniper EX and QFX series switches, VSTP is restricted to a maximum of 253 or 510 VLAN instances depending on the software version (ELS vs. non-ELS). In your scenario, having 450 VLANs exceeds the standard 253-instance limit found on many platforms.

The Solution (Option B): When the number of VLANs exceeds the VSTP capacity, the recommended best practice is to enable RSTP (Rapid Spanning Tree Protocol). Unlike VSTP, RSTP runs a single spanning tree instance for the entire switch, regardless of how many VLANs are configured. This ensures that all 450 VLANs are protected from loops without hitting the hardware ' s instance-count limit.

Other Options: Option A (force-version) only affects the BPDU format for compatibility but doesn ' t solve the instance limit. Option C and Option D are parameter tuning actions that do not address the architectural limitation of the number of running instances.