Introduction

Welcome to the comprehensive VXLAN Fabric Technology guide. This guide takes you from Layer 2 fundamentals through advanced VXLAN fabric architectures, preparing you for both implementation and interviews.

The Evolution of Data Center Networking

Spine 1
Spine 2
Spine 3
| | | | | | | | | |
Leaf 1
Leaf 2
Leaf 3
Leaf 4
| | | | | | | |
Server
Server
Server
Server

Part I: Layer 2 Fundamentals

Understanding the foundation before the evolution

1 Understanding VLANs

Chapter 1: What is a VLAN?

Definition: A VLAN (Virtual Local Area Network) is a logical grouping of network devices that allows them to communicate as if they were on the same physical network segment, regardless of their physical location.

✓ When to Use VLANs

  • Segmenting broadcast domains
  • Grouping devices by department/function
  • Implementing security policies
  • Small to medium networks (<4000 segments)
  • Single data center deployments

✗ When NOT to Use VLANs

  • Need more than 4094 segments
  • Multi-datacenter connectivity
  • Need VM mobility across locations
  • Large-scale cloud environments
  • Use VXLAN instead for these cases

Why VLANs Were Created

Without VLANs (Problem)

  • Single broadcast domain
  • All devices receive broadcasts
  • Network congestion
  • Security concerns
  • No logical separation

With VLANs (Solution)

  • Separate broadcast domains
  • Traffic isolated by VLAN
  • Reduced congestion
  • Enhanced security
  • Logical grouping achieved

VLAN Port Types

Port Type Description Use Case
Access Port Belongs to single VLAN End devices (PCs, printers)
Trunk Port Carries multiple VLANs Switch-to-switch connections
Native VLAN Untagged VLAN on trunk Management, backward compatibility

VLAN Fundamentals Diagram

VLAN Fundamentals VLAN 10 (Sales) VLAN 20 (Engineering) Layer 2 Switch PC-A VLAN 10 10.1.10.10 PC-B VLAN 10 10.1.10.20 PC-C VLAN 20 10.1.20.10 PC-D VLAN 20 10.1.20.20 Gi0/1 Gi0/2 Gi0/3 Gi0/4 Port Configuration Port | Mode | VLAN | Status Gi0/1 | Access | 10 | Connected (PC-A) Gi0/2 | Access | 10 | Connected (PC-B) Gi0/3 | Access | 20 | Connected (PC-C) Legend: VLAN 10 (Sales) VLAN 20 (Engineering) Devices in same VLAN can communicate

Figure 1.1: VLAN network topology with port assignments and VLAN membership

802.1Q VLAN Tagging

Standard Ethernet Frame: +------------+------------+------+---------+-----+ | Dest MAC | Src MAC | Type | Payload | FCS | | (6 bytes) | (6 bytes) |(2B) | |(4B) | +------------+------------+------+---------+-----+ 802.1Q Tagged Frame: +------------+------------+--------+------+---------+-----+ | Dest MAC | Src MAC |802.1Q | Type | Payload | FCS | | (6 bytes) | (6 bytes) |Tag(4B) |(2B) | |(4B) | +------------+------------+--------+------+---------+-----+ | v TPID: 0x8100 | VID: 1-4094

Configuration Examples

Cisco IOS

! Create VLANs
Switch(config)# vlan 10
Switch(config-vlan)# name Sales

! Configure Access Port
Switch(config)# interface GigabitEthernet0/1
Switch(config-if)# switchport mode access
Switch(config-if)# switchport access vlan 10

! Configure Trunk Port
Switch(config)# interface GigabitEthernet0/24
Switch(config-if)# switchport mode trunk
Switch(config-if)# switchport trunk allowed vlan 10,20

Juniper Junos

# Create VLANs
set vlans Sales vlan-id 10

# Configure Access Port
set interfaces ge-0/0/1 unit 0 family ethernet-switching vlan members Sales

# Configure Trunk Port
set interfaces ge-0/0/24 unit 0 family ethernet-switching port-mode trunk
set interfaces ge-0/0/24 unit 0 family ethernet-switching vlan members [Sales Engineering]

VLAN Limitations

Limitation Impact Modern Solution
4,094 VLAN limit Cannot scale for multi-tenant DCs VXLAN (16M VNIs)
Spanning Tree required Blocks redundant links VXLAN + ECMP
Layer 2 boundary VLANs don't cross routers VXLAN overlay
2 The CAM Table

Chapter 2: How the CAM Table is Built

Definition: The CAM (Content Addressable Memory) Table, also known as the MAC Address Table, is a database maintained by switches that maps MAC addresses to physical ports.

The Learning Process

MAC Address Learning Animation

Frame arrives on Port 1 with Source MAC: AA:AA:AA:AA:AA:AA

Switch learns: MAC AA:AA -> Port 1

Step-by-Step Process

  1. Frame Reception: Frame arrives on a switch port
  2. Source MAC Learning: Switch records source MAC + ingress port
  3. Destination Lookup: Switch checks if destination MAC is in table
  4. Forward or Flood: Known = forward to port; Unknown = flood to all ports in VLAN
  5. Timer Management: Entries age out (default 300 seconds)

CAM Table Learning Process

Step 1: Frame Arrives Switch CAM: Empty PC-A AA:AA Frame Step 2: Learn Source MAC Switch CAM: AA:AA -> P1 Source MAC learned! Step 3: Destination Unknown = Flood Switch Dest: BB:BB NOT FOUND P2 P3 P4 Flood to all ports in VLAN Final CAM Table MAC Address | Port | VLAN | Timer AA:AA:AA:AA:AA:AA | P1 | 10 | 300s BB:BB:BB:BB:BB:BB | P2 | 10 | 295s CC:CC:CC:CC:CC:CC | P3 | 20 | 180s

Figure 2.1: CAM table learning process showing frame arrival, MAC learning, and flooding

CAM Table Example

MAC Address Port VLAN Timer
AA:AA:AA:AA:AA:AA Port 1 10 300s
BB:BB:BB:BB:BB:BB Port 2 10 245s
CC:CC:CC:CC:CC:CC Port 3 20 180s

Verification Commands

! Cisco
show mac address-table
show mac address-table vlan 10
show mac address-table count

! Juniper
show ethernet-switching table
show ethernet-switching table vlan Sales
3 First Hop Redundancy

Chapter 3: First Hop Redundancy Protocols

Definition: First Hop Redundancy Protocols (FHRP) provide default gateway redundancy by allowing multiple routers to share a virtual IP address.

FHRP Comparison

Protocol Full Name Standard Virtual MAC
HSRP Hot Standby Router Protocol Cisco Proprietary 0000.0c07.acXX
VRRP Virtual Router Redundancy Protocol RFC 5798 0000.5e00.01XX
GLBP Gateway Load Balancing Protocol Cisco Proprietary 0007.b400.XXYY
In VXLAN Fabrics: Traditional FHRP is replaced by Distributed Anycast Gateway, where all leaves share the same gateway IP and MAC - no failover needed!

Part II: Evolution to VXLAN

From traditional VLANs to modern overlay networks

4 VXLAN Fundamentals

Chapter 4: VXLAN Fundamentals

Definition: VXLAN (Virtual Extensible LAN) is an encapsulation protocol that creates virtualized Layer 2 networks over Layer 3 infrastructure, enabling Layer 2 connectivity across Layer 3 boundaries.

✓ When to Use VXLAN

  • Large-scale environments (>4000 segments)
  • Multi-datacenter deployments
  • VM mobility requirements
  • Cloud and SDN environments
  • Need for multi-tenancy
  • Active-Active redundancy requirements

✗ When NOT to Use VXLAN

  • Small networks with <100 segments
  • Legacy hardware that doesn't support it
  • Simple flat network requirements
  • No Layer 3 infrastructure available
  • MTU constraints that can't be resolved

Key VXLAN Components

Component Description Role
VNI VXLAN Network Identifier 24-bit ID (16 million unique values)
VTEP VXLAN Tunnel Endpoint Encapsulates/decapsulates frames
NVE Network Virtualization Edge Logical interface for VXLAN
Underlay Physical IP network Transports encapsulated packets
Overlay Virtual Layer 2 network Logical network seen by endpoints

VXLAN Encapsulation

Original Ethernet Frame: +----------+----------+------+---------+-----+ | Dst MAC | Src MAC | Type | Payload | FCS | +----------+----------+------+---------+-----+ After VXLAN Encapsulation: +-------+--------+-----+-------+-------------------------+ | Outer | Outer | UDP | VXLAN | Original Ethernet Frame | | L2 | IP |4789 | VNI | | +-------+--------+-----+-------+-------------------------+ | | | | | +-- 24-bit VNI (16 million segments) | +-- VTEP IP addresses +-- Next-hop MAC addresses

VXLAN Encapsulation Format

Original Ethernet Frame: Dst MAC Src MAC Type Payload FCS VXLAN Encapsulation VXLAN Encapsulated Packet: Outer L2 Outer IP UDP 4789 VXLAN VNI Original Ethernet Frame VTEP IPs 24-bit VNI (16M segments) VLAN vs VXLAN Comparison Feature | VLAN | VXLAN Segment ID Size | 12 bits (4K) | 24 bits (16M) Transport | Layer 2 | Layer 3 (UDP) Key Points - UDP Port 4789 - 50+ byte overhead - Requires MTU 9000+ - Supports 16M segments

Figure 4.1: VXLAN encapsulation showing packet header structure and comparison with VLAN

5 VLAN vs VXLAN

Chapter 5: VLAN vs VXLAN Comparison

Feature VLAN VXLAN
Segment ID Size 12 bits (4,094) 24 bits (16,777,216)
Transport Layer 2 Layer 3 (UDP)
Spanning Tree Required Not needed
Redundancy Active/Standby (STP) Active/Active (ECMP)
Scalability Limited Massive
Control Plane None (data plane) BGP EVPN (optional)

VLAN Architecture (3-Tier)

  • STP blocks 50% of links
  • Oversubscription at aggregation
  • Variable latency
  • Complex scaling

VXLAN Architecture (Leaf-Spine)

  • All links active (ECMP)
  • Predictable 2-hop latency
  • Non-blocking
  • Linear scaling

Part III: VXLAN Fabric Architecture

Understanding Clos, Underlay, and Overlay

6 Clos Architecture

Chapter 6: Clos Architecture

Definition: Clos architecture is a multi-stage, non-blocking network topology that provides full-bandwidth connectivity between any two endpoints. Named after Charles Clos (Bell Labs, 1953).

Leaf-Spine Clos Architecture

Spine 1
Spine 2
Spine 3
Spine 4
Every leaf connects to every spine (ECMP)
Leaf 1
VTEP
Leaf 2
VTEP
Leaf 3
VTEP
Leaf 4
VTEP
Servers connect to leaves
Servers
Servers
Servers
Servers

Leaf-Spine Clos Architecture with Interface Labels

SPINE LAYER LEAF LAYER (VTEPs) SERVER LAYER Spine 1 10.0.0.1 Spine 2 10.0.0.2 Spine 3 10.0.0.3 Spine 4 10.0.0.4 Leaf 1 VTEP 10.0.1.1 Leaf 2 VTEP 10.0.1.2 Leaf 3 VTEP 10.0.1.3 Leaf 4 VTEP 10.0.1.4 Leaf 5 VTEP 10.0.1.5 Servers Servers Servers Servers Servers Eth1/49 Eth1/50 Clos Architecture Properties Non-blocking: Full bandwidth between any endpoints Predictable: Consistent 2-hop latency (Leaf-Spine-Leaf) ECMP: All links active, load balanced Scalable: Add spines for bandwidth, leaves for ports Traffic Path: Leaf1 -> Any Spine -> Leaf3 (Always 2 hops with ECMP)

Figure 6.1: Clos leaf-spine architecture with VTEPs and interface numbering

Clos Properties

Property Description
Non-blocking Full bandwidth between any endpoints
Predictable Consistent 2-hop path (leaf-spine-leaf)
ECMP Equal-Cost Multi-Path load balancing
Scalable Add spines for bandwidth, leaves for ports

Clos vs MAC Addresses

Aspect Clos Architecture MAC Addresses
Purpose Network topology design Device identification
Layer Physical + Layer 3 Layer 2
Function Traffic path selection Frame forwarding
Learning Static (configured) Dynamic (CAM table)
7-8 Underlay vs Overlay

Chapter 7-8: Underlay and Overlay Networks

Aspect Underlay Overlay
Definition Physical IP infrastructure Virtual network on top
Layer Layer 3 (IP) Layer 2 (logical)
Protocol OSPF, IS-IS, BGP VXLAN
Addressing VTEP IP addresses Endpoint MAC addresses
Purpose Transport VXLAN packets Provide tenant connectivity
+------------------------------------------------------------------+ | OVERLAY NETWORK | | (Virtual Layer 2 - VNI 10000) | | [VM-A] ---------------------------------------- [VM-B] | +------------------------------------------------------------------+ | VXLAN Encapsulation | +------------------------------------------------------------------+ | UNDERLAY NETWORK | | (Physical Layer 3 - IP) | | VTEP-1 Spine 1 Spine 2 VTEP-2 | | 10.0.1.1 -----> 10.0.0.1 <--> 10.0.0.2 <----- 10.0.1.2 | +------------------------------------------------------------------+

Underlay vs Overlay Networks

OVERLAY NETWORK (Virtual Layer 2 - VNI 10000) VM-A VLAN 100 VM-B VLAN 100 Logical L2 Connection VXLAN Encapsulation UNDERLAY NETWORK (Physical Layer 3 - IP) VTEP-1 10.0.1.1 Spine 1 10.0.0.1 Spine 2 10.0.0.2 VTEP-2 10.0.1.2 Comparison Aspect | Underlay | Overlay Layer | Layer 3 (IP) | Layer 2 (logical) Protocol | OSPF/BGP/IS-IS | VXLAN Purpose | Transport | Tenant connectivity

Figure 7.1: Relationship between underlay physical network and overlay virtual network

Part IV: Control Plane - BGP EVPN

The brain of the VXLAN fabric

9-10 BGP EVPN

Chapter 9-10: BGP EVPN and Route Types

Definition: BGP EVPN (Ethernet VPN) is a control plane protocol that distributes MAC and IP reachability information across the VXLAN fabric, eliminating the need for flood-and-learn.

EVPN Route Types

Type Name Purpose
Type-1 Ethernet Auto-Discovery Multi-homing, fast convergence
Type-2 MAC/IP Advertisement MAC and IP reachability (most used)
Type-3 Inclusive Multicast BUM traffic handling
Type-4 Ethernet Segment Designated Forwarder election
Type-5 IP Prefix External IP routing

BGP EVPN Configuration (Cisco NX-OS)

! Enable BGP EVPN
nv overlay evpn

router bgp 65000
  router-id 10.0.1.1
  neighbor 10.0.0.1 remote-as 65000
    update-source loopback0
    address-family l2vpn evpn
      send-community extended

evpn
  vni 10000 l2
    rd auto
    route-target import auto
    route-target export auto

BGP EVPN Control Plane

Route Reflector (Spine / RR) Leaf 1 VTEP 10.0.1.1 Leaf 2 VTEP 10.0.1.2 Leaf 3 VTEP 10.0.1.3 iBGP EVPN Host A MAC-A 10.100.1.10 Host B MAC-B 10.100.1.20 Host C MAC-C 10.100.2.10 Eth1/1 Eth1/1 Eth1/1 EVPN Route Types Type | Name | Purpose Type-1 | Ethernet Auto-Discovery | Multi-homing Type-2 | MAC/IP Advertisement | MAC reachability Type-3 | Inclusive Multicast | BUM handling Type-5 | IP Prefix | External routing MAC Learning via EVPN: 1. Host A connects to Leaf 1 2. Leaf 1 learns MAC-A locally 3. Leaf 1 sends Type-2 route via BGP 4. All leaves learn: MAC-A -> VTEP 10.0.1.1 5. No flooding required!

Figure 9.1: BGP EVPN control plane with route reflector and route type distribution

Part V: Traffic Flow in VXLAN

Understanding how packets traverse the fabric

11-13 Traffic Flow

Chapter 11-13: Intra-VXLAN, Inter-VXLAN, and BUM Traffic

Traffic Types in VXLAN

Traffic Type Description VNI Used
Intra-VXLAN (Same VNI) Hosts in same subnet L2VNI (e.g., 10000)
Inter-VXLAN (Different VNI) Hosts in different subnets L3VNI (e.g., 50000)
BUM Traffic Broadcast, Unknown unicast, Multicast L2VNI with replication

BUM Handling Methods

Multicast Underlay

Map VNIs to multicast groups. Requires PIM in underlay.

  • Efficient delivery
  • Requires multicast infrastructure

Ingress Replication

Source VTEP sends unicast copy to each remote VTEP.

  • No multicast needed
  • More CPU at source

BGP EVPN (Recommended)

EVPN Type-3 routes advertise VTEP membership.

  • Scalable and dynamic
  • Most common today

VXLAN Traffic Flow (Same VNI)

Server A 10.100.1.10 VLAN 100 Leaf 1 (VTEP-1) 10.0.1.1 Encapsulate Spine Layer (ECMP) Leaf 2 (VTEP-2) 10.0.1.2 Decapsulate Server B 10.100.1.20 VLAN 100 eth0 Eth1/1 Eth1/49-50 Eth1/1-48 Eth1/1-48 Eth1/49-50 Eth1/1 eth0 Step 1 Original L2 Step 2 VNI 10000 Step 3 IP Routed Step 4 Decapsulated VXLAN Packet Details: - Outer Src IP: 10.0.1.1 (VTEP-1) - Outer Dst IP: 10.0.1.2 (VTEP-2) - UDP Port: 4789 - VNI: 10000 BUM Traffic Handling: Multicast VNI -> mcast group Efficient delivery Ingress Replication Unicast to each VTEP No mcast needed BGP EVPN (Recommended) Type-3 routes Scalable, dynamic

Figure 11.1: VXLAN traffic flow showing encapsulation, routing through spines, and decapsulation

Part VI: Multicast in VXLAN

Multicast group strategies and considerations

14-15 Multicast Groups

Chapter 14-15: Multicast Groups and VNIs

Multicast Group Assignment Strategies

Strategy Description When to Use
One Group per VNI VNI 10000 -> 239.1.1.1
VNI 10001 -> 239.1.1.2		 Small deployments (<100 VNIs)
Shared Groups VNI 10000-10099 -> 239.1.1.1
VNI 10100-10199 -> 239.1.1.2		 Large deployments (100+ VNIs)
Ingress Replication No multicast groups No multicast in underlay / BGP EVPN
Best Practice: With BGP EVPN, use Ingress Replication. It's scalable, doesn't require multicast infrastructure, and dynamically learns VTEP membership.

Part VII: Vendor Implementations

Cisco DNA vs Juniper Campus Fabric

16-17 Vendor Comparison

Chapter 16-17: Cisco DNA and Juniper Campus Fabric

Comparison Table

Aspect Cisco SD-Access Juniper Campus Fabric
Control Plane LISP BGP EVPN
Data Plane VXLAN-GPO VXLAN
Management DNA Center (mandatory) Mist Cloud / Junos Space
Underlay IS-IS (auto) eBGP or OSPF
Segmentation SGT (TrustSec) Micro-segmentation
Hardware Catalyst 9000 QFX/EX Series

Cisco DNA Center Communication

API/Protocol Use Case
REST APIs Automation, integration, provisioning
NETCONF/YANG Configuration management
SSH/CLI Device access
SNMP Monitoring, polling

Cisco SD-Access vs Juniper Campus Fabric

Cisco SD-Access Control Plane: LISP Data Plane: VXLAN-GPO Management: DNA Center (required) Underlay: IS-IS (auto-provisioned) Segmentation: SGT / TrustSec Hardware: Catalyst 9000 series CISCO Juniper Campus Fabric Control Plane: BGP EVPN Data Plane: VXLAN Management: Mist Cloud / Junos Space Underlay: eBGP or OSPF Segmentation: Micro-segmentation Hardware: QFX/EX Series JUNIPER Key Differences LISP vs EVPN: Different control plane philosophies DNA Center is mandatory for Cisco; Juniper is more flexible Cisco is more prescriptive; Juniper is more modular

Figure 16.1: Side-by-side comparison of Cisco SD-Access and Juniper Campus Fabric architectures

Part VIII: External Connectivity

Fusion routers and traffic leaving the fabric

18-19 External Traffic

Chapter 18-19: Fusion Router and External Traffic

VNI Extension Options

Option 1: VXLAN to Fusion

Fusion router acts as VTEP and participates in BGP EVPN.

  • Full VXLAN integration
  • Seamless routing

Option 2: VLAN Handoff

Border leaf terminates VXLAN and trunks VLANs to fusion.

  • Legacy compatibility
  • Simpler fusion router

Automation Capabilities

Task Can Automate? Method
VNI Creation Yes DNA Center / Ansible
VNI-to-VLAN Mapping Yes Templates
BGP EVPN Peering Partial ZTP + Templates
External Routing Manual Security review needed

Part IX: Hands-On Labs

Practice configurations for Cisco and Juniper

Lab Exercises

Lab 1: Basic VXLAN Configuration (Cisco)

Objective: Configure basic VXLAN with multicast BUM handling

! Enable features
feature nv overlay
feature vn-segment-vlan-based

! Create VLAN and map to VNI
vlan 100
  name Production
  vn-segment 10000

! NVE Interface
interface nve1
  no shutdown
  source-interface loopback0
  member vni 10000
    mcast-group 239.1.1.1

Lab 2: BGP EVPN Setup (Cisco)

Objective: Add BGP EVPN control plane

! Enable BGP EVPN
nv overlay evpn

router bgp 65000
  neighbor 10.0.0.1 remote-as 65000
    address-family l2vpn evpn
      send-community extended

evpn
  vni 10000 l2
    rd auto
    route-target import auto
    route-target export auto

Lab 3: VXLAN with Juniper

Objective: Configure EVPN-VXLAN on Junos

# VLAN and VNI
set vlans Production vlan-id 100
set vlans Production vxlan vni 10000

# VTEP source
set switch-options vtep-source-interface lo0.0
set switch-options route-distinguisher 10.0.1.1:1
set switch-options vrf-target target:65000:1

# EVPN
set protocols evpn encapsulation vxlan
set protocols evpn extended-vni-list 10000

Lab 4: Verification Commands

! Cisco
show nve peers
show nve vni
show bgp l2vpn evpn summary
show bgp l2vpn evpn route-type 2

! Juniper
show ethernet-switching vxlan-tunnel-end-point remote
show evpn database
show route table bgp.evpn.0

Part X: Interview Preparation

Questions from basic to consultant level

Interview Questions & Answers

Basic Level

Q1: What is a VLAN?

A VLAN (Virtual Local Area Network) is a logical grouping of network devices that allows them to communicate as if they were on the same physical network segment. VLANs create separate broadcast domains within a switch.

Key points:

  • VLAN IDs: 1-4094
  • 802.1Q standard for tagging
  • Access ports (single VLAN) vs Trunk ports (multiple VLANs)
Q2: How does a switch build its MAC address table?

Through source MAC learning:

  1. Frame arrives on a port
  2. Switch records source MAC + port
  3. Timer starts (default 300 seconds)
  4. Entry refreshed on each frame
  5. Aged out if timer expires

For forwarding: Known MAC = specific port; Unknown = flood

Q3: What is VXLAN and why was it developed?

VXLAN (Virtual Extensible LAN) creates Layer 2 overlay networks over Layer 3 infrastructure.

Why developed:

  • VLANs limited to 4,094; VXLAN has 16 million VNIs
  • STP blocks links; VXLAN uses ECMP
  • VLANs can't cross L3; VXLAN can
  • Cloud scale and multi-tenancy requirements

Intermediate Level

Q4: Explain BGP EVPN and its route types.

BGP EVPN distributes MAC and IP reachability information across VXLAN fabrics.

Route Types:

  • Type-1: Ethernet Auto-Discovery (multi-homing)
  • Type-2: MAC/IP Advertisement (most common)
  • Type-3: Inclusive Multicast (BUM handling)
  • Type-4: Ethernet Segment (DF election)
  • Type-5: IP Prefix (external routing)
Q5: How does traffic flow between two hosts in different VXLANs?

Inter-VXLAN routing using Symmetric IRB:

  1. Source sends to anycast gateway on local leaf
  2. Leaf routes in VRF, determines destination subnet
  3. Encapsulates with L3VNI for tenant VRF
  4. Routes through spines to destination leaf
  5. Destination leaf decapsulates L3VNI
  6. Routes to destination L2VNI
  7. Forwards to destination host

Key: L2VNI for bridging, L3VNI for routing between subnets

Advanced/Consultant Level

Q6: Compare Cisco SD-Access with Juniper Campus Fabric.

Key Differences:

  • Control Plane: Cisco uses LISP; Juniper uses BGP EVPN
  • Management: DNA Center is mandatory for Cisco; Juniper is flexible
  • Segmentation: Cisco uses SGT/TrustSec; Juniper uses micro-segmentation
  • Underlay: Cisco auto-provisions IS-IS; Juniper uses eBGP or OSPF

Cisco is more prescriptive; Juniper is more modular.

Q7: How would you troubleshoot VXLAN connectivity issues?

Systematic approach:

  1. Underlay: Check OSPF/BGP neighbors, ping VTEP IPs, verify MTU (9000+)
  2. VXLAN Config: Verify NVE interface, VNI mappings
  3. Control Plane: Check BGP EVPN neighbors and routes
  4. NVE Peers: Verify peer state and VNIs
  5. MAC Learning: Check local and remote MAC entries

Common issues: MTU too small, underlay not advertising loopbacks, NVE source mismatch

Q8: Design a VXLAN fabric for 10,000 endpoints.

Key Design Decisions:

  • Spine Layer: 4 spines with 48x 100G ports each
  • Leaf Layer: 13+ leaves (800 endpoints per leaf)
  • Underlay: eBGP (unique AS per leaf)
  • Overlay: VXLAN with BGP EVPN, ingress replication
  • Gateway: Anycast gateway on all leaves
  • HA: vPC/MC-LAG at leaf, ECMP across spines
  • External: Dual border leaves with firewall integration

Glossary

Term Definition
ARPAddress Resolution Protocol - maps IP to MAC
BGPBorder Gateway Protocol - routing protocol
BUMBroadcast, Unknown unicast, Multicast traffic
CAMContent Addressable Memory (MAC table)
ClosMulti-stage non-blocking network topology
ECMPEqual-Cost Multi-Path load balancing
EVPNEthernet VPN - BGP extension for L2/L3
IRBIntegrated Routing and Bridging
VNIVXLAN Network Identifier
VTEPVXLAN Tunnel Endpoint
VRFVirtual Routing and Forwarding