IPv6 Addressing in AVD

Introduction

AVD provides a powerful automatic IPv6 addressing system that assigns IPv6 addresses to fabric devices based on pools and node identifiers. This eliminates manual IPv6 planning and ensures consistent, predictable addressing across your network.

This guide covers IPv6 pool types, allocation mechanisms, and customization options available in AVD for pure IPv6 numbered underlay deployments.

RFC 5549 Not Covered

This guide focuses on IPv6 numbered addressing. RFC 5549 (IPv6 link-local with IPv4 NLRI) uses a different approach and is not covered here.

Key Concepts

Before diving into specifics, understand these core concepts:

IPv6 Pool: A range of IPv6 addresses from which AVD allocates individual addresses

Node ID: A unique numeric identifier for each device, used to calculate address offsets

Offset: A value added to the base pool address to derive a specific IPv6 address

Prefix Length: The subnet mask size for allocated addresses (e.g., /128, /64)

Enabling IPv6 Addressing

To use IPv6 pools for underlay addressing, enable IPv6 numbered mode at the fabric level:

Enable IPv6 Addressing

---
# How-To IPv6 Fabric Settings

# Override to maintain Management0 as the management interface for cEOS
custom_platform_settings:
  - platforms: [ceos]
    management_interface: Management0

default_interfaces:
  - types: [spine]
    platforms: [default]
    uplink_interfaces: [Ethernet1-2]
    downlink_interfaces: [Ethernet1-8]
  - types: [l3leaf]
    platforms: [default]
    uplink_interfaces: [Ethernet1-2]
    mlag_interfaces: [Ethernet3-4]

default_node_types:
  - match_hostnames:
      - ".*-spine.*"
    node_type: spine
  - match_hostnames:
      - ".*-leaf.*"
    node_type: l3leaf

# Enable IPv6 underlay with numbered addresses
underlay_ipv6: true
underlay_ipv6_numbered: true

# IPv6 prefix length settings
fabric_ip_addressing:
  loopback:
    ipv6_prefix_length: 128
  mlag:
    ipv6_prefix_length: 64
  p2p_uplinks:
    ipv6_prefix_length: 64

# Updating BGP peer group names for IPv6
bgp_peer_groups:
  ipv4_underlay_peers:
    name: IPv6-UNDERLAY-PEERS
  mlag_ipv4_underlay_peer:
    name: MLAG-IPv6-UNDERLAY-PEER

Requirements

IPv6 numbered underlay requires:

• underlay_ipv6: true - Enables IPv6 on underlay interfaces
• underlay_ipv6_numbered: true - Uses explicit IPv6 addresses (not link-local)
• loopback_ipv6_pool defined for each node type
• router_id_pool for BGP Router ID (IPv4 format required by BGP)

IPv6 Pool Types

AVD uses several IPv6 pools for different purposes:

Pool Variable	Purpose	Default Interface
loopback_ipv6_pool	Router ID and BGP peering	Loopback0
vtep_loopback_ipv6_pool	VXLAN tunnel endpoints	Loopback1
uplink_ipv6_pool	P2P links between devices	Ethernet uplinks
mlag_peer_ipv6_pool	MLAG peer-link SVI	VLAN 4094
mlag_peer_l3_ipv6_pool	MLAG L3 iBGP peering	VLAN 4093
router_id_pool	BGP Router ID only (IPv4)	None (ID only)

Pool Hierarchy

Pools can be defined at multiple levels with the following precedence (highest to lowest):

1. Node level - Specific to a single device
2. Node group level - Shared by devices in a group (e.g., MLAG pair)
3. Node type defaults - Applied to all nodes of a type

Router ID Pool

Even in pure IPv6 deployments, BGP requires an IPv4-formatted Router ID. The router_id_pool provides this without configuring any IPv4 addresses on interfaces.

Loopback IPv6 Allocation

Basic Formula

For Loopback0 (Router ID interface):

IPv6 = loopback_ipv6_pool + nodes[].id + loopback_ipv6_offset

Note

When spines and leafs share the same pool, use loopback_ipv6_offset to prevent address conflicts. The offset is 0 by default.

Spine Example

Spine assignment from pool 2001:db8:1::/48:

• spine1 (id=1): 2001:db8:1::1/128 - calculation (pool + id[1])
• spine2 (id=2): 2001:db8:1::2/128 - calculation (pool + id[2])

Loopback0 for spines

---
# How-To IPv6 Spine Configuration

spine:
  defaults:
    platform: ceos
    # IPv4 pool for BGP Router ID (required for IPv6 numbered underlay)
    router_id_pool: 10.255.0.0/27
    # IPv6 pool for Loopback0 addresses
    loopback_ipv6_pool: 2001:db8:1::/48
    bgp_as: 65100
  nodes:
    - name: htipv6-numbered-spine1
      id: 1
      mgmt_ip: 172.16.3.11/24
    - name: htipv6-numbered-spine2
      id: 2
      mgmt_ip: 172.16.3.12/24

spine1 configuration

interface Loopback0
   description ROUTER_ID
   no shutdown
   ipv6 address 2001:db8:1::1/128
!

spine2 configuration

interface Loopback0
   description ROUTER_ID
   no shutdown
   ipv6 address 2001:db8:1::2/128
!

Leaf Example

Leaf assignment from pool 2001:db8:1::/48 with offset of 2:

• leaf1 (id=1): 2001:db8:1::3/128 - calculation (pool + id[1] + offset[2])
• leaf2 (id=2): 2001:db8:1::4/128 - calculation (pool + id[2] + offset[2])

Loopback0 for leafs

---
# How-To IPv6 Leaf Configuration

l3leaf:
  defaults:
    platform: ceos
    # IPv4 pool for BGP Router ID (required for IPv6 numbered underlay)
    router_id_pool: 10.255.1.0/27
    # IPv6 pool for Loopback0 addresses
    loopback_ipv6_pool: 2001:db8:1::/48
    loopback_ipv6_offset: 2
    # IPv6 pool for VTEP Loopback1 addresses
    vtep_loopback_ipv6_pool: 2001:db8:3::/48
    # IPv6 pool for P2P uplinks
    uplink_ipv6_pool: 2001:db8:10::/48
    uplink_switches: [htipv6-numbered-spine1, htipv6-numbered-spine2]
    # MLAG peer address family set to IPv6
    mlag_peer_address_family: ipv6
    # IPv6 pool for MLAG peer-link SVI
    mlag_peer_ipv6_pool: 2001:db8:20::/48
    # IPv6 pool for MLAG L3 iBGP peering
    mlag_peer_l3_ipv6_pool: 2001:db8:21::/48
    virtual_router_mac_address: 00:1c:73:00:00:99
    spanning_tree_priority: 4096
    spanning_tree_mode: mstp

  node_groups:
    - group: HTIPV6_L3_LEAFS
      bgp_as: 65101
      nodes:
        - name: htipv6-numbered-leaf1
          id: 1
          mgmt_ip: 172.16.3.101/24
          uplink_switch_interfaces: [Ethernet1, Ethernet1]
        - name: htipv6-numbered-leaf2
          id: 2
          mgmt_ip: 172.16.3.102/24
          uplink_switch_interfaces: [Ethernet2, Ethernet2]

leaf1 loopback configuration

interface Loopback0
   description ROUTER_ID
   no shutdown
   ipv6 address 2001:db8:1::3/128
!
interface Loopback1
   description VXLAN_TUNNEL_SOURCE
   no shutdown
   ipv6 address 2001:db8:3:3::1/128
!

leaf2 loopback configuration

interface Loopback0
   description ROUTER_ID
   no shutdown
   ipv6 address 2001:db8:1::4/128
!
interface Loopback1
   description VXLAN_TUNNEL_SOURCE
   no shutdown
   ipv6 address 2001:db8:3:3::1/128
!

MLAG VTEP Sharing

MLAG pairs share the same VTEP IPv6 address. AVD automatically uses the MLAG primary ID in the group for both peers. This translates to IPv6 = pool + mlag_primary_id + loopback_ipv6_offset. Non-MLAG nodes use their own ID.

P2P Uplink IPv6 Allocation

Uplink IPv6 addresses are calculated using a formula to ensure unique /64 subnets for each link.

Formula

subnet_offset = ([nodes[].id - 1] * max_uplink_switches * max_parallel_uplinks) + uplink_switch_index

Where:

• nodes[].id: The leaf’s ID
• max_uplink_switches: Maximum number of uplink switches (default: length of uplink_switches)
• max_parallel_uplinks: Maximum parallel uplinks per switch (default: 1)
• uplink_switch_index: Index of the uplink switch (0-based)

Example

P2P Uplink for leafs

---
# How-To IPv6 Leaf Configuration

l3leaf:
  defaults:
    platform: ceos
    # IPv4 pool for BGP Router ID (required for IPv6 numbered underlay)
    router_id_pool: 10.255.1.0/27
    # IPv6 pool for Loopback0 addresses
    loopback_ipv6_pool: 2001:db8:1::/48
    loopback_ipv6_offset: 2
    # IPv6 pool for VTEP Loopback1 addresses
    vtep_loopback_ipv6_pool: 2001:db8:3::/48
    # IPv6 pool for P2P uplinks
    uplink_ipv6_pool: 2001:db8:10::/48
    uplink_switches: [htipv6-numbered-spine1, htipv6-numbered-spine2]
    # MLAG peer address family set to IPv6
    mlag_peer_address_family: ipv6
    # IPv6 pool for MLAG peer-link SVI
    mlag_peer_ipv6_pool: 2001:db8:20::/48
    # IPv6 pool for MLAG L3 iBGP peering
    mlag_peer_l3_ipv6_pool: 2001:db8:21::/48
    virtual_router_mac_address: 00:1c:73:00:00:99
    spanning_tree_priority: 4096
    spanning_tree_mode: mstp

  node_groups:
    - group: HTIPV6_L3_LEAFS
      bgp_as: 65101
      nodes:
        - name: htipv6-numbered-leaf1
          id: 1
          mgmt_ip: 172.16.3.101/24
          uplink_switch_interfaces: [Ethernet1, Ethernet1]
        - name: htipv6-numbered-leaf2
          id: 2
          mgmt_ip: 172.16.3.102/24
          uplink_switch_interfaces: [Ethernet2, Ethernet2]

Resulting allocations for leaf1 (id=1) from pool 2001:db8:10::/48:

Uplink	Subnet Offset	Leaf IPv6	Spine IPv6
To spine1	0	2001:db8:10::2/64	2001:db8:10::1/64
To spine2	1	2001:db8:10:1::2/64	2001:db8:10:1::1/64

leaf1 P2P to spines

interface Ethernet1
   description P2P_htipv6-numbered-spine1_Ethernet1
   no shutdown
   mtu 9214
   no switchport
   ipv6 address 2001:db8:10::2/64
!
interface Ethernet2
   description P2P_htipv6-numbered-spine2_Ethernet1
   no shutdown
   mtu 9214
   no switchport
   ipv6 address 2001:db8:10:1::2/64
!

MLAG IPv6 Allocation

MLAG requires two pools for peer connectivity:

Pool	Purpose	Default VLAN
mlag_peer_ipv6_pool	L2 peer-link SVI	4094
mlag_peer_l3_ipv6_pool	L3 iBGP peering	4093

Enabling MLAG IPv6

To use IPv6 for MLAG peering, set mlag_peer_address_family to ipv6:

MLAG IPv6 Settings

---
# How-To IPv6 Leaf Configuration

l3leaf:
  defaults:
    platform: ceos
    # IPv4 pool for BGP Router ID (required for IPv6 numbered underlay)
    router_id_pool: 10.255.1.0/27
    # IPv6 pool for Loopback0 addresses
    loopback_ipv6_pool: 2001:db8:1::/48
    loopback_ipv6_offset: 2
    # IPv6 pool for VTEP Loopback1 addresses
    vtep_loopback_ipv6_pool: 2001:db8:3::/48
    # IPv6 pool for P2P uplinks
    uplink_ipv6_pool: 2001:db8:10::/48
    uplink_switches: [htipv6-numbered-spine1, htipv6-numbered-spine2]
    # MLAG peer address family set to IPv6
    mlag_peer_address_family: ipv6
    # IPv6 pool for MLAG peer-link SVI
    mlag_peer_ipv6_pool: 2001:db8:20::/48
    # IPv6 pool for MLAG L3 iBGP peering
    mlag_peer_l3_ipv6_pool: 2001:db8:21::/48
    virtual_router_mac_address: 00:1c:73:00:00:99
    spanning_tree_priority: 4096
    spanning_tree_mode: mstp

  node_groups:
    - group: HTIPV6_L3_LEAFS
      bgp_as: 65101
      nodes:
        - name: htipv6-numbered-leaf1
          id: 1
          mgmt_ip: 172.16.3.101/24
          uplink_switch_interfaces: [Ethernet1, Ethernet1]
        - name: htipv6-numbered-leaf2
          id: 2
          mgmt_ip: 172.16.3.102/24
          uplink_switch_interfaces: [Ethernet2, Ethernet2]

MLAG Allocation Algorithms

AVD supports three MLAG IPv6 allocation algorithms configured via fabric_ip_addressing.mlag.algorithm:

first_id (default)odd_idsame_subnet

Uses the first node’s ID in the MLAG group:

fabric_ip_addressing:
  mlag:
    algorithm: first_id  # Default

Formula: offset = (mlag_primary_id - 1)

Requires one node with an odd ID and one with an even ID:

fabric_ip_addressing:
  mlag:
    algorithm: odd_id

Formula: offset = (odd_id - 1)

All MLAG pairs share the same subnet (first in pool):

fabric_ip_addressing:
  mlag:
    algorithm: same_subnet

Formula: offset = 0 (always)

leaf1 MLAG VLANs

vlan 4093
   name MLAG_L3
   trunk group MLAG
!
vlan 4094
   name MLAG
   trunk group MLAG
!
interface Vlan4093
   description MLAG_L3
   no shutdown
   mtu 9214
   ipv6 address 2001:db8:21::1/64
!
interface Vlan4094
   description MLAG
   no shutdown
   mtu 9214
   no autostate
   ipv6 address 2001:db8:20::1/64
!

Global IPv6 Addressing Settings

Configure fabric-wide IPv6 addressing behavior:

IPv6 Addressing Settings

fabric_ip_addressing:
  mlag:
    algorithm: first_id  # first_id, odd_id, same_subnet
    ipv6_prefix_length: 64
  p2p_uplinks:
    ipv6_prefix_length: 64
  loopback:
    ipv6_prefix_length: 128

Setting	Default	Valid Values	Description
loopback.ipv6_prefix_length	128	64, 128	Loopback interface prefix length
mlag.ipv6_prefix_length	64	1-127	MLAG peer-link and L3 peering prefix length
p2p_uplinks.ipv6_prefix_length	64	1-127	P2P uplink prefix length

Pool Formats

AVD supports flexible IPv6 pool formats:

Single Prefix

loopback_ipv6_pool: 2001:db8:1::/48

Multiple Prefixes

Comma-separated list of prefixes:

loopback_ipv6_pool: 2001:db8:1::/56, 2001:db8:2::/56

IPv6 Ranges

loopback_ipv6_pool: 2001:db8::1-2001:db8::ff

Static IPv6 Overrides

Override VTEP loopback with a static value:

l3leaf:
  nodes:
    - name: htipv6-numbered-leaf1
      id: 1
      vtep_loopback_ipv6_address: 2001:db8:101::1 # Override VTEP pool

Override Variable	Overrides Pool
vtep_loopback_ipv6_address	vtep_loopback_ipv6_pool

Note

There is no loopback_ipv6_address override for Loopback0. Use loopback_ipv6_pool with loopback_ipv6_offset to control IPv6 address allocation.

Unsupported Features

Some AVD features are not yet supported with IPv6 numbered underlay:

• underlay_multicast_pim_sm
• underlay_multicast_rp_interfaces
• underlay_rfc5549 (mutually exclusive with numbered)
• wan_role
• vtep_vvtep_ip
• inband_ztp

Best Practices

1. Plan your ID scheme: Use consistent ID numbering across the fabric
2. Use offsets wisely: When sharing pools between node types, use loopback_ipv6_offset
3. Size pools appropriately: Ensure pools have enough addresses for growth
4. Use /128 for loopbacks: Prevents routing table bloat
5. Use /64 for P2P links: AVD defaults to /64 prefix length for point-to-point links, though /127 can be configured per RFC 6164
6. Always define router_id_pool: BGP requires IPv4 Router ID even in IPv6-only deployments
7. Use node groups for MLAG: Define MLAG pairs as node groups for automatic VTEP sharing

Troubleshooting

Common Issues

Issue	Cause	Solution
BGP session not establishing	Missing router_id_pool	Define IPv4 pool for Router ID
No IPv6 addresses on interfaces	underlay_ipv6 not enabled	Set underlay_ipv6: true
MLAG using IPv4	mlag_peer_address_family not set	Set to ipv6
Validation error on routing protocol	Wrong underlay protocol	Ensure underlay_routing_protocol is ebgp (default for spine/l3leaf)
Missing VTEP IPv6 address	vtep_loopback_ipv6_pool not defined	Define pool for leaf nodes
IPv6 address conflict between spine and leaf	Shared pool without offset	Add loopback_ipv6_offset to leaf defaults
Pool exhausted	Too many nodes for pool size	Use larger pool or multiple prefixes

Reference

• Node Type VTEP and Loopback Configuration
• Fabric IP Addressing
• MLAG Configuration