You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Support L3 capabilities in connected endpoints. AVD currently only supports L2 for connected endpoints. This is beneficial for both M&E and AI deployments.
Which component of AVD is impacted
eos_designs
Use case example
M&E and AI deployments currently use following configs for L3 endpoints,
For M&E fabric,
interface Ethernet{{ interface }}
description {{ description }}
no shutdown
mtu {{mtu number}}
no switchport
ip address {{ ipv4_address }}/{{prefix /30 or /31 or /29}}
ptp enable
ptp sync-message interval -3
ptp announce interval 0
ptp transport ipv4
ptp announce timeout 3
ptp delay-req interval -3
ptp role {{roleName}}
multicast ipv4 static or pim ipv4 sparse-mode
no error-correction encoding (should not be default config)
ip helper-address 10.246.32.8 # DHCP ip helper
ip helper-address 10.247.32.8
ip igmp last-member-query-count 0
speed
For AI Fabric, following sub config is also beneficial,
service-profile QOS-CPU-PROFILE"
Describe the solution you would like
Ability to set routed endpoints . ( It may be useful to lump them all in one file to support large deployments)
Ability to mention this for a range of interfaces
Ability to set this for range of interfaces and also able to group them if possible
example: EtX-Y is all for EndpointA with dedicated pool of ip address and qos profile
Ability to allocate a pool of ip addresses and distribute it as /31 or /30 or /29.
Ability to handle breakouts.
Optional: Being able to pull endpoint data (interface description, ip address, etc) from Netbox
Describe alternatives you have considered
Using profiles and separate yml files in host_vars for every switch having connected endpoint.
Enhancement summary
Support L3 capabilities in connected endpoints. AVD currently only supports L2 for connected endpoints. This is beneficial for both M&E and AI deployments.
Which component of AVD is impacted
eos_designs
Use case example
M&E and AI deployments currently use following configs for L3 endpoints,
For M&E fabric,
interface Ethernet{{ interface }}
description {{ description }}
no shutdown
mtu {{mtu number}}
no switchport
ip address {{ ipv4_address }}/{{prefix /30 or /31 or /29}}
ptp enable
ptp sync-message interval -3
ptp announce interval 0
ptp transport ipv4
ptp announce timeout 3
ptp delay-req interval -3
ptp role {{roleName}}
multicast ipv4 static or pim ipv4 sparse-mode
no error-correction encoding (should not be default config)
ip helper-address 10.246.32.8 # DHCP ip helper
ip helper-address 10.247.32.8
ip igmp last-member-query-count 0
speed
For AI Fabric, following sub config is also beneficial,
service-profile QOS-CPU-PROFILE"
Describe the solution you would like
example: EtX-Y is all for EndpointA with dedicated pool of ip address and qos profile
Describe alternatives you have considered
Using profiles and separate yml files in host_vars for every switch having connected endpoint.
l3_edge:
p2p_links_ip_pools:
- name: RedLeaf1_pool
ipv4_pool: 10.91.196.0/24
prefix_size: 30
p2p_links_profiles:
- name: HostProfile
mtu: 1500
ptp_enable: true
raw_eos_cli: |
ptp role master
multicast ipv4 static
p2p_links:
Additional context
No response
Contributing Guide
The text was updated successfully, but these errors were encountered: