Help us improve your experience.

Let us know what you think.

Do you have time for a two-minute survey?

 
ON THIS PAGE
 

Easy EVPN LAG (EZ-LAG) Configuration

SUMMARY With the EZ-LAG configuration feature, you can easily configure a small Ethernet virtual private network (EVPN) for a pair of peer provider edge (PE) devices that have attached multihomed or single-homed servers. You use a simplified Junos OS CLI statement hierarchy, and a built-in commit script generates the full configuration.

EVPN fabric PE devices reliably handle traffic to and from attached multihomed end devices by grouping the multihoming links into an EVPN Ethernet segment with an identifier (ESI). You configure the links participating in the Ethernet segment into a link aggregation group (LAG), so we call the set of multihomed links an ESI LAG. In this document, we call the PE devices peer PE devices when they link to multihomed end devices with the same ESI. The PE devices might also have one or more links in a LAG to single-homed end devices, although an EVPN PE device doesn't need to handle single-homed links as an Ethernet segment.

The end devices might be hosts or servers directly attached to the PE devices, or might be customer edge (CE) devices with attached end hosts or servers. For simplicity, in this document we refer to the end devices collectively as servers.

EVPN Multihoming provides redundancy and load balancing between the two switches, and provides a loop-free Layer 2 (L2) network without running STP. However, configuring EVPN multihoming can be complex, and involves configuring many statements correctly across the PE devices in the fabric. With this easy EVPN LAG configuration feature (also called the EZ-LAG feature), we provide a simplified configuration statement hierarchy and a built-in commit script that you can use to set up EVPN multihoming. This feature makes it easy to migrate a multichassis link aggregation group (MC-LAG) topology to a standards-based EVPN-VXLAN multihoming model.

See Figure 1. The supported EVPN topology includes:

  • Two peer PE devices connected back-to-back with an aggregated Ethernet interface bundle

  • An EVPN fabric with VXLAN encapsulation interconnecting the peer PE devices

  • ESI LAG configurations between the peer PE devices and one or more attached multihomed servers

  • Connections between a peer PE device and one or more attached single-homed servers

Figure 1: Setups Supported for Easy EVPN LAG Configuration Setups Supported for Easy EVPN LAG Configuration

To use this feature, you need either of the following software licenses:

  • Advanced 1 for EZ-LAG (Single EVPN Peer)

  • Advanced 2 for EVPN-VXLAN

Refer to the Juniper Licensing User Guide for details on the Juniper Flex Software License model and Juniper Agile Licensing.

Benefits of Using Easy EVPN LAG Configuration

  • Automated configuration: You can easily set up a small EVPN fabric, including ESI-LAGs for multihomed attached servers, without using a network controller. You provide some essential parameters through the CLI, and the device's built-in commit script transforms them into a complete EVPN fabric configuration.

  • Configuration flexibility: You have the flexibility to use the basic and default configuration options, and also customize the generated configuration. You can override many of the commit script default behaviors, and manually configure those elements.

  • Simplified topology migration: You can use this feature to easily migrate a multichassis link aggregation group (MC-LAG) topology to a simple EVPN-VXLAN fabric.

Easy EVPN LAG Configuration Overview

The easy EVPN LAG configuration feature operates at configuration commit time. It consists of:

  • A set of configuration statements at the [edit services evpn] hierarchy level with which you provide the parameters to set up the prescribed EVPN fabric.

  • A built-in commit script that processes the simplified configuration elements at commit time (before the standard Junos OS configuration validity checks) and generates a corresponding EVPN fabric configuration.

You configure a few [edit services evpn] statements to provide the minimal, required set of parameters to configure the EVPN fabric and the links to the attached servers.

When you commit the [edit services evpn] configuration statements, the device invokes the commit script that generates the corresponding EVPN configuration using the parameters you provided. The commit script derives some configuration statements specific to the device on which you commit the simplified configuration. The commit process validates the configuration and generates warning messages for missing required parameters or misconfigured parameters.

See Commit Script Overview and How Commit Scripts Work for more on commit scripts.

At any time, you can easily include a few more [edit services evpn] statements to add elements to the existing configuration, such as new VLANs or connections to new servers.

Overview of the [edit services evpn] Statement Hierarchy

You can specify the minimal set of parameters in just a few configuration statements. For example, consider the following small EVPN-VXLAN fabric with two connected multihomed servers that each host two VLANs:

Figure 2: Small Example EVPN Topology Small Example EVPN Topology

The following few easy EVPN LAG configuration commands provide the minimal required parameters to configure the EVPN fabric from Figure 2 on device Peer PE 1. When you commit these configuration commands, the commit script generates a corresponding full default configuration that might contain 80 to 100 configuration commands:

See Easy EVPN LAG Configuration with Multihomed Servers for the full configuration the commit script generates.

The [edit services evpn] command hierarchy has many additional options that give you flexibility to customize the default generated configuration as needed. The commit script uses some of the elements you provide in the simplified configuration to automatically derive other parameters in the generated configuration. By default, the commit script also generates configuration commands for other common features in an EVPN fabric, such as loop detection and storm control on the interfaces from PE devices to servers.

You can include options to:

  • Tell the commit script not to automatically derive some parameters in the generated configuration.

  • Override some of the commit script default values and behaviors.

  • Apply configuration groups to statements the commit script generates.

With any of those options, if the commit script requires any such elements to provide a valid configuration, you must configure those elements manually.

See Simplified CLI Statements and Parameters to Generate the Configuration for details on the parameters each statement and option provide to the commit script, and how those parameters affect the generated configuration.

Built-in Commit Script

The commit script for this feature, services_evpn_commit_script.py, is enabled by default on supported platforms.

Simplified CLI Statements and Parameters to Generate the Configuration

This section gives details about the set of configuration statements at the [edit services evpn] hierarchy level. We also show how the commit script maps the simplified configuration parameters to generate the full configuration.

Default Generated Configuration Overview

The default generated configuration includes:

  • Underlay and overlay peering using external BGP (EBGP).

  • A default MAC-VRF EVPN instance named __SERVICES_EVPN_EVPN_VXLAN_MAC-VRF_1, with VLAN-aware service type and VXLAN encapsulation, which hosts the VLANs you specify.

    You can also configure additional MAC-VRF instances and their member VLANs.

  • IRB interfaces with the specified IPv4 (or IPv6) addresses for routing between VLANs.

    By default, the commit script derives a virtual gateway address for each IRB interface as the highest configurable address in the specified IRB subnet address range.

    The commit script also assigns a default virtual gateway MAC address of 00:00:5e:00:01:01 if you don't specify that parameter.​

  • ESI LAG connections to multihomed servers, and LAG links to single-homed servers.

    The commit script assigns aggregated Ethernet interfaces to those links as trunk interfaces with VLAN tagging, flexible Ethernet services encapsulation, and LACP enabled.

  • Lightweight loop detection on the server-facing aggregated Ethernet interfaces, with loop detection action interface down. See EVPN-VXLAN Lightweight Leaf to Server Loop Detection for more on this feature.

  • Storm control on server-facing interfaces.

    The commit script generates a default storm control profile named __SERVICES_EVPN_EVPN_VXLAN_STORM_CONTROL, and assigns it to the each server-facing interface. See MAC Filtering, Storm Control, and Port Mirroring Support in an EVPN-VXLAN Environment for more on the storm control feature.

See an example easy LAG configuration and the corresponding generated configuration here: Easy EVPN LAG Configuration with Multihomed Servers.

The next sections describe the configuration elements in detail.

Easy EVPN LAG Configuration Statements and Options

Table 1 shows the statements and options at the [edit services evpn] configuration statement hierarchy level. The commit script requires a minimum set of statements to have enough information to generate a working configuration. The table indicates which statements are mandatory. You can optionally include other statements to change the commit script behavior and override default values as needed for your topology and configuration preferences.

The commit script uses some of the elements you provide in the simplified configuration to automatically derive other parameters in the generated configuration. By default, the commit script also generates configuration statements for supporting common features in an EVPN fabric, such as loop detection and storm control on links to servers.

Table 1: [edit services evpn] Statements and Purpose
Statement and Child Statements Options Purpose Mandatory?

defaults-override

If you configure any of these options, the script will not automatically derive those parameters from the related [edit services evpn] statements you provided, or it will skip generating configuration for the related element. You must manually configure those elements instead.

 

no-aggregate-device-count-config

Don't generate the default aggregated Ethernet (aen) interface count configuration statement.

No

no-loop-detect-config

Don't generate lightweight loop detection configuration statements.

No

no-platform-defaults-config

Don't generate statements for default platform-specific options.

No

no-policy-and-routing-options-config

Don't generate default routing and policy options configuration statements.

If you specify this option, you must manually provide configuration for a policy-statement named EXPORT-LO0 at the [edit policy-options policy-statement] hierarchy level. The commit script assigns a policy by that name in the default generated EBGP underlay configuration.

Or, you must include the no-underlay-config option to override generating the default underlay configuration, and manually configure the underlay peering statements you want instead.

No

no-storm-control-config

Don't generate configuration for storm control.

No

no-overlay-bgp-config

Don't generate BGP configuration for the overlay.

No

no-underlay-config

Don't generate configuration for the underlay peering.

No

device-attribute

With these statements, you define parameters to configure the peer PE devices, including the peer interconnection interfaces, underlay peering, and overlay peering between them to form the EVPN fabric. The commit script configures underlay and overlay peering with external BGP (EBGP) by default. You can optionally specify that the commit script generate configuration for one of the other available underlay or overlay protocols. Or, you can alternatively set the no-overlay-bgp-config option or the no-underlay-config option (at the [edit services evpn defaults-override] hierarchy level), and manually configure the overlay BGP peering statements or underlay peering statements you want.

 

peer-id peer-id

Number (1-2) that identifies the PE device (compared to its peer PE device) in the fabric.

The commit script uses this value to derive some values that must be unique across the peer PE devices in the generated configuration.

Yes

system-id system-id

Peer PE device's system ID (MAC address format).

The commit script needs this value to derive configuration to set up LACP on ESI LAG interfaces.

Yes

(device-attribute)

loopback

peer1-subnet peer1-subnet

peer2-subnet peer2-subnet

PE device loopback IPv4 subnet addresses for peer PE 1 and peer PE 2

From this parameter, the commit script configures the loopback interface subnet addresses for the peer PE devices, and the router ID for each device.

Yes

(device-attribute)

peer-to-peer

peer-subnet (inet | inet6) subnet-address

IPv4 or IPv6 subnet address for the interfaces that connect the PE device to its peer PE device.

You can specify this option with either the inet or the inet6 option, or both.

Yes

peer-subnet interface-name [ interface-name … ]

Name or names of the interfaces that connect the PE device to its peer PE device.

Yes

overlay-connectivity { ibgp }

Generate a prescribed configuration for overlay peering using internal BGP (IBGP) instead of the default protocol, EBGP.

No

underlay-connectivity { ospf }

Generate a prescribed configuration for underlay peering using OSPF instead of the default protocol, EBGP.

No

evpn-vxlan

With these statements, you define parameters to configure the peer PE devices to run EVPN with VXLAN encapsulation using one or more VLAN-aware MAC-VRF instances.

By default, the commit script enables EVPN-VXLAN in a single MAC-VRF instance named __SERVICES_EVPN_EVPN_VXLAN_MAC_VRF_1. If you configure the mac-vrf-instance instance-id option at the [edit services evpn expn-vxlan irb irb-instance instance] hierarchy level, the commit script generates configuration using that instance-id instead. The commit script appends the instance-id to the base string to configure unique MAC-VRF instance names in the generated configuration: __SERVICES_EVPN_EVPN_VXLAN_MAC_VRF_instance-id.

If needed, you can configure more than one MAC-VRF instance for tenant traffic separation at L2.

These statements also provide the parameters to set up the ESI-LAGs to multihomed servers or LAGs to single-homed servers, including the IRB interfaces and the associated VLANs. You can optionally specify to configure DHCP relay for the IRB instances under this stanza as well.

dhcp-relay

name

DHCP relay group name.

No

dhcp-server-address dhcp-server-address

IP address of the DHCP server to enable the DHCP server relay group.

No

relay-source DHCP-relay-source-interface

IP address of the DHCP relay source loopback interface.

The commit script generates DHCP overrides relay-source statements that use this source address.

No

vrf-instance vrf-instance-id

VRF instance identifier in this configuration for which you want to configure DHCP relay.

If you don't specify a vrf-instance-id, the commit script uses the default VRF instance (inet.0 routing table). Otherwise, the commit script generates configuration for a VRF instance named __SERVICES_EVPN_EVPN_VXLAN_VRF_vrf-instance-id.

No

irb

irb-instance

String to identify an IRB instance and its associated parameters for commit script processing.

We recommend choosing IDs that match the parameters of the IRB instance in the generated configuration. For example, use "irb_10" for the IRB instance associated with VLAN 10, which will be irb.10 in the generated configuration.

Yes

apply-config-groups config-groups

Apply the specified configuration groups to the configuration the commit script generates from the statements at this hierarchy level.

No

no-dhcp-relay

Exclude this IRB instance from the DHCP relay configuration.

This option applies only if you specified to generate configuration for DHCP relay with the dhcp-relay option at the [edit services evpn expn-vxlan] hierarchy level.

No

use-anycast-address

Configure the IRB instance with an anycast gateway address.

With this option, you must include the anycast-mac mac-address at the [edit services evpn global-parameters] hierarchy level to specify the anycast MAC address to use.

No

vlan-id vlan-num

VLAN ID associated with the IRB instance.

The commit script derives the IRB interface name with a matching logical unit number. For example, the commit script configures irb.10 for vlan-num 10.

Yes

(irb irb-instance)

instance

mac-vrf-instance instance-id

The MAC-VRF instance ID to which the IRB instance belongs.

The commit script generates MAC-VRF instance configuration statements using MAC-VRF instance name __SERVICES_EVPN_EVPN_VXLAN_MAC_VRF_instance-id.

No*

*If you don't configure this option, the commit script creates a MAC-VRF instance using default instance-id 1.

vrf-instance instance-id

The Layer 3 (L3) virtual routing and forwarding (VRF) instance to which this IRB instance belongs.

If you specify this option, the commit script generates VRF instance configuration statements for VRF instance name __SERVICES_EVPN_EVPN_VXLAN_VRF_instance-id.

No*

*If you don't configure this option, the commit script uses the default VRF instance (which corresponds to the inet.0 routing table).

(irb irb-instance)

subnet-address

inet (ipv4-subnet-address | [ipv4-addr1 ipv4-addr 2 …])

IPv4 subnet address or list of subnet addresses for the IRB interface.

For simplicity, you can configure the same IRB subnet address on both peer PE devices. By default, the commit script uses the PE peer ID to derive different IRB subnet addresses for each peer device based on the configured ipv4-subnet-address value.

Alternatively, you can set the subnet addresses you want on each peer PE device, and include the no-irb-address-auto-derive option so the commit script uses those exact addresses.

Also, see subnet-address for details on additional default router-advertisement configuration statements the commit script generates for the IRB interfaces.

Yes*

*You must configure at least one subnet address for the IRB instance using either the inet option or the inet6 option.

inet6 (ipv6-subnet-address | [ipv6-addr1 ipv6-addr 2 …])

IPv6 subnet address for the IRB interface.

The commit script has the same default behavior with this option as with the inet option above. You can similarly include the no-irb-address-auto-derive option and specify the exact IRB interface IPv6 subnet addresses you want on each device. Also, see subnet-address for details on the additional router-advertisement configuration statements the commit script generates by default when you specify a link-local address with this inet6 option.

Yes*

*You must configure at least one subnet address for the IRB instance using either the inet option or the inet6 option.

no-irb-address-auto-derive

Don't derive IRB subnet addresses in the generated configuration for this IRB instance.

You must specify this option if you don't want the commit script default behavior that uses the PE peer ID to derive different IRB subnet addresses for each peer device from the same configured (inet | inet6) address.

With this option, the commit script uses the exact inet and inet6 subnet addresses you provide.

This option is also available at the global-parameters hierarchy level that applies to all IRB instances.

No

virtual-gateway-v4-address virtual-gateway-v4-address

Virtual gateway IPv4 address for the IRB interface.

If you don't include this option, by default the commit script derives a virtual gateway IPv4 address from the inet ipv4-subnet-address value. The commit script uses the highest configurable IPv4 address in that subnet range. For example, subnet address 10.1.1.1/24 generates virtual gateway address 10.1.1.254.

No

virtual-gateway-v6-address virtual-gateway-v6-address

Virtual gateway IPv6 address for the IRB interface.

If you don't include this option, the commit script has the same default behavior to derive the virtual-gateway-v6-address as it does for the virtual-gateway-v4-address above. For example, IPv6 subnet address 2001:db8::10:1:1:1/112 generates virtual gateway IPv6 address 2001:db8::10:1:1:fffe.

No

server

name

A unique name to identify a server connected to the peer PE device, such as SERVER_1 or HostA.

This value is used internally to associate server-related parameters with the statements to generate for that server. You might see this server name in description parameters for related statements in the generated configuration.

Yes

apply-config-groups config-groups

Apply the specified configuration groups to the configuration the commit script generates from the statements at this hierarchy level.

No

enable-pxe-boot

Enable this connected server to use a preboot execution environment (PXE) boot process.

The commit script configures LACP force-up status on the link to the server to ensure the link is UP when the server boots up. The PXE boot process requires this status.

No

esi-lag-id esi-lag-id

ESI LAG connection ID for this server.

The commit script assigns an aggregated Ethernet interface for the connection to this server by adding esi-lag-id to the base interface name ae0 (so for example, esi-lag-id 1 uses ae1).

Yes*

*You must specify at least one esi-lag-id OR single-home-id for each server. See the single-home-id option in the next row.)

single-home-id single-home-id

Single-homed connection ID for the named server.

The commit script assigns an aggregated Ethernet interface for the connection to this server by adding single-home-id to the base interface name ae1024 (so for example, single-home-id 1 uses ae1025).

Yes*

*You must specify at least one esi-lag-id OR single-home-id for each server.

(See the esi-lag-id option in the previous row.)

interface (interface-name | [ interface-name …])

Interface name or list of interface names for the physical links from the peer PE device to this server.

Yes

vlan-id-list [vlan-id-list]

List of VLAN IDs hosted by this MAC-VRF instance and server.

Yes*

*You must configure at least one VLAN ID list (with at least one VLAN ID in it) for each server, either at this server name level or at the mac-vrf-instance instance-id level (see next row).

(server name)

mac-vrf-instance

instance-id

Identifier for a MAC-VRF instance that hosts this server.

If you don't specify this option, the commit script generates configuration that associates these server parameters with the default MAC-VRF instance (named __SERVICES_EVPN_EVPN_VXLAN_MAC_VRF_1).

No

vlan-id-list [vlan-id-list]

List of VLAN IDs hosted by this MAC-VRF instance and server.

Yes*

*You must configure at least one VLAN ID list (with at least one VLAN ID in it) for each server, either at this MAC-VRF instance level or at the server name level.

global-parameters

With these statements, you specify parameters for configuration elements that are common across the peer PE devices in the EVPN fabric, such as the default VLAN and anycast gateway addresses. The commit script has default values it uses if you don't configure these options.
 

anycast-mac anycast-mac

Globally use the specified anycast virtual gateway MAC address in the generated configuration.

No

default-vlan vlan-id

Use this VLAN ID (1-4094) as the default VLAN in the generated configuration.

The commit script uses the system default VLAN ID if you don't configure this.

No

no-irb-address-auto-derive

Don't derive IRB interface subnet addresses in the generated configuration for all IRB instances. Instead, use the exact IPv4 or IPv6 subnet addresses you specify with the subnet-address statement at the [edit services evpn evpn-vxlan irb irb-instance] hierarchy level.

Otherwise, by default, the commit script derives the IRB subnet address from the subnet-address (inet | inet6) statement based on the PE device peer-id.

You can alternatively set this option at the irb irb-instance hierarchy level to apply only to a specific IRB instance.

No

start-aggregate-ethernet-index num

Use this number as the starting index in aggregated Ethernet interface names for ESI LAG links to attached servers.

The commit script starts at ae1 by default for ESI LAG links to servers, and reserves ae0 interface names only for the links between the peer PEs.

Note:

This option does not affect generated interface names for links to single-homed servers. The commit script uses aggregated Ethernet interface names starting at ae1025 for single-homed server links.

No
virtual-gateway-mac virtual-gateway-mac

Use this virtual gateway MAC address for both IPv4 and IPv6 traffic in the generated configuration.

Specify the virtual-gateway option (later in this table) instead of this option if you want to assign different virtual gateway MAC addresses for IPv4 and IPv6 traffic.

No

(global-parameters)

mtu

overlay overlay-mtu

Specify the maximum transmission unit (MTU) in bytes to set in the generated configuration for the interfaces used in the overlay peering, instead of the default MTU value.

The default MTU value varies by platform and type of interface (or protocol) for which you configure the value.

No

underlay underlay-mtu

Specify the MTU in bytes to set in the generated configuration for the interfaces used in the overlay peering, instead of the default MTU value.

The default MTU value varies by platform and type of interface (or protocol) for which you configure the value.

No

(global-parameters)

virtual-gateway

v4-mac v4-mac

Globally use this virtual gateway MAC address for IPv4 traffic in the generated configuration.

The commit script uses 00:00:5e:00:01:01 by default if you don't configure this option or the virtual-gateway-mac option (above).

No

v6-mac v6-mac

Globally use this virtual gateway MAC address for IPv6 traffic in the generated configuration.

The commit script uses 00:00:5e:00:02:01 by default if you don't configure this option or the virtual-gateway-mac option (see that row earlier in this table).

No

How the Commit Script Uses Easy EVPN LAG Configuration Elements in the Generated Configuration

Table 2 lists configuration elements and default values that the commit script uses or derives from the simplified configuration you provide.

Table 2: Derived and Default Configuration Values
Configuration Element Default or Derived Value

General or Shared Elements

Number of aggregated Ethernet interface connected device for set chassis aggregated-devices ethernet device-count num in generated configuration

num is 255 by default.

(Some platforms might use a different default platform-specific value.)

Peer PE identifier, to associate specified or derived parameters with the corresponding peer PE device

From set services evpn device-attribute peer-id peer-id.

Parameter peer-id is required for each peer PE device (there is no default value).

The peer-id can have value 1 or 2.

Peer PE device loopback interface (lo0) address

From set services evpn device-attribute:

  • peer-id peer-id

  • loopback peer1-subnet peer1-subnet or loopback peer2-subnet peer2-subnet

Use the loopback interface address on peer PE device peer-id from the provided loopback subnet addess for that peer ID.

Router ID, for set routing-options router-id router-id in generated configuration

From set services evpn device-attribute:

  • peer-id peer-id

  • loopback peer1-subnet peer1-subnet or loopback peer2-subnet peer2-subnet

Assigned the same value as the device's loopback interface address.

Load balancing routing options policy for per-flow load balancing

 
set policy-options policy-statement pplb then load-balance per-packet
set routing-options forwarding-table export pplb

Storm control configuration for each server-facing interface

 
set forwarding-options storm-control-profiles __SERVICES_EVPN_EVPN_VXLAN_STORM_CONTROL all bandwidth-percentage 1
set interfaces interface-name ether-options ethernet-switch-profile storm-control __SERVICES_EVPN_EVPN_VXLAN_STORM_CONTROL

Peer to Peer PE Device Connections

You specify a peer PE device ID used internally to track the parameters that apply to that configured EVPN peer PE device as follows:

set services evpn device-attribute peer-id peer-id

Aggregated Ethernet interface name

ae0

Aggregated Ethernet interface address

From set services evpn device-attribute:

For configuration simplicity, you configure the same subnet-address on both peer PE devices.

The commit script derives the interface address based on peer-id as follows:

  • peer-id 1: ae0 address is subnet-address.

  • Otherwise, ae0 address is subnet-address + 1 (add 1 to the low-order address segment of the subnet range).

For example, if the IPv4 subnet-address is 10.0.1.0/31:

  • On peer-id 1, ae0 = 10.0.1.0

  • Otherwise (on peer-id 2) ae0 = 10.0.1.1

Aggregated Ethernet interface member physical interface or interfaces

From set services evpn device-attribute peer-to-peer peer-subnet interface-name [ interface-name ... ].

 Parameter [ interface-name ... ] is required; set those interfaces as member links in ae0.

Peer PE Device to Server Connections

You specify server names and assign IDs that represent the links to those servers (either multihomed ESI LAG connections or single-homed connections), as follows:

  • set services evpn evpn-vxlan server server-name:

    • esi-lag-id esi-lag-id

    • single-home-id single-home-id

The commit script uses the server names internally to associate configured elements with an attached server, but also displays these names in the description element of related generated configuration statements.
Starting index for assigning aggregated Ethernet interface names for ESI LAG links to multihomed servers

Default is 1 (ae1)

To override the default value and assign a different starting index (num), configure:

set service evpn global-parameters start-aggregate-ethernet-index num.

This option affects only the starting index for aggregated Ethernet interfaces to multihomed servers (esi-lag-id option). The commit script uses ae(num + esi-lag-id).

This option doesn't affect how the commit script allocates single-homed server links (single-home-id option), which is ae(1024 + single-home-id).

Name of physical interface or interfaces from peer PE device to server (single interface or list of interfaces)

From set services evpn evpn-vxlan server server-name interface [ interface-name ... ]

Required parameter per server, no default

Aggregated Ethernet interface name (aeindex) and logical unit (unit num)

From set services evpn evpn-vxlan server server-name:

  • esi-lag-id esi-lag-id or single-home-id single-home-id

  • mac-vrf-instance instance-id
 Derive as follows based on server link ID and configured MAC-VRF instance ID (or default MAC-VRF instance-id 1):

  • For multihomed servers: ae(esi-lag-id) unit instance-id

  • For single-homed servers: ae(1024 + single-home-id) unit instance-id

Aggregated Ethernet interface options and member VLANs

From set services evpn evpn-vxlan server server-name:

  • esi-lag-id esi-lag-id or single-home-id single-home-id

  • mac-vrf-instance instance-id

  • vlan-id-list [ vlan-id ...]

Enable VLAN tagging, flexible Ethernet services encapsulation, and trunk interface mode.

Set member VLANs from configured server vlan-id-list

For example, for esi-lag-id 1 (interface name ae1) in the default MAC-VRFinstance-id 1 with vlan-id-list [ 10 20], the generated configuration is:

set interfaces ae1 vlan-tagging 
set interfaces ae1 encapsulation flexible-ethernet-services 
set interfaces ae1 unit 1 family ethernet-switching interface-mode trunk
set interfaces ae1 unit 1 family ethernet-switching vlan members 10
set interfaces ae1 unit 1 family ethernet-switching vlan members 20

System ID for LACP on PE device to server links (MAC address format)

From:

  • set services evpn device-attribute​ system-id system-id

  • set services evpn evpn-vxlan server server-name:

    • esi-lag-id esi-lag-id

    • single-home-id single-home-id

Derive from system-id per server link by adding the server link ID as follows:

system-id + esi-lag-id

OR

system-id + single-home-id

For example, if system-id is 10:11:12:13:14:15, then for esi-lag-id 1 (ae1), LACP system ID is 10:11:12:13:14:16 in the generated LACP configuration for that link:

set interfaces ae1 aggregated-ether-options lacp active
set interfaces ae1 aggregated-ether-options lacp system-id 10:11:12:13:14:16
Note:

To skip configuring LACP for a peer-server link, include the following option:

set services evpn evpn-vxlan server server-name no-lacp

Multihomed server interface ESI LAG configuration

From set services evpn evpn-vxlan server server-name esi-lag-id esi-lag-id

Automatically derive the ESI:

set interfaces aeesi-lag-id esi auto-derive type-1-lacp
set interfaces aeesi-lag-id all-active

Llightweight loop detection configuration on server-facing aggregated Ethernet interfaces

From set services evpn evpn-vxlan server server-name:

  • esi-lag-id esi-lag-id or single-home-id single-home-id

  • vlan-id-list [ vlan-id ...]

 
set protocols loop-detect enhanced interface aeesi-lag-id vlan-id vlan-id
set protocols loop-detect enhanced interface aeesi-lag-id loop-detect-action interface-down 
set protocols loop-detect enhanced interface aeesi-lag-id transmit-interval 1s
set protocols loop-detect enhanced interface aeesi-lag-id revert-interval 60

For more on this loop detection feature, see EVPN-VXLAN Lightweight Leaf to Server Loop Detection.

EVPN, VXLAN, and IRB Interface Elements

Supported EVPN instance parameters in generated configuration: MAC-VRF instance type, VLAN-aware service type, and VXLAN encapsulation

MAC-VRF EVPN instance name

From:

  • set services evpn evpn-vxlan irb irb-instance instance mac-vrf-instance instance-id.

  • set services evpn evpn-vxlan server server-name mac-vrf-instance instance-id

If you don't configure one or more mac-vrf-instance instance-id statements:

  • The default MAC-VRF instance instance-id is 1.

  • The default MAC-VRF instance name is __SERVICES_EVPN_EVPN_VXLAN_MAC_VRF_1.

Otherwise, for each configured instance-id, the MAC-VRF instance name is __SERVICES_EVPN_EVPN_VXLAN_MAC_VRF_instance-id.

Aggregated Ethernet interface logical unit per MAC-VRF instance

From set services evpn evpn-vxlan server server-name:

  • mac-vrf-instance instance-id

  • esi-lag-id esi-lag-id or single-home-id single-home-id

 Derive and configure associated logical interface as follows:

set routing-instances __SERVICES_EVPN_EVPN_VXLAN_MAC_VRF_instance-id interface ae(esi-lag-id | single-home-id).instance-id

Route distinguisher (RD) for EVPN instance

From :

  • set services evpn device-attribute:

    • peer-id peer-id

    • loopback peer1-subnet peer1-subnet

    • loopback peer2-subnet peer2-subnet

  • set services evpn evpn-vxlan server server-name mac-vrf-instance instance-id

Route distinguisher for each MAC-VRF routing instance is derived from the peer PE device loopback subnet address and the MAC-VRF instance-id as follows:

peer<peer-id>-subnet:instance-id

Target extended community for EVPN instance

From mac-vrf-instance instance-id

vrf-target:1:instance-id

VLANs (bridge domains) hosted in the EVPN instance

From:

  • set services evpn evpn-vxlan server server-name vlan-id-list [vlan-id ...]

  • set services evpn evpn-vxlan irb irb-instance vlan-id vlan-id

VLAN names are SERVICES_EVPN_EVPN_VXLAN_VLAN_vlan-id

IRB interface logical units corresponding to configured server VLANs

From:

  • set services evpn evpn-vxlan irb irb-instance vlan-id vlan-id

  • set services evpn evpn-vxlan server server-name vlan-id-list [vlan-id ...]

IRB interface logical unit is vlan-id in generated configuration for:

  • set interfaces irb unit vlan-id family inet address subnet-address ...

  • set routing-instances __SERVICES_EVPN_EVPN_VXLAN_MAC_VRF_instance-id vlans SERVICES_EVPN_EVPN_VXLAN_VLAN_vlan-id l3-interface irb.vlan-id​

IRB interface address (to configure set interfaces irb unit unit family inet address subnet-address ...)

From set services evpn evpn-vxlan irb irb-instance subnet-address (inet | inet6) subnet-address

If you want the generated configuration to have different IRB interface subnet-address values that you explicitly set per peer PE device (instead of the default addresses derived from a common configured subnet-address value), then configure the following option:

  • Per-IRB instance:

    set services evpn evpn-vxlan irb irb-instance no-irb-instance-auto-derive

  • Globally (for all IRB instances):

    set services evpn global-parameters no-irb-instance-auto-derive

Derive different addresses per peer PE device from IPv4 or IPv6 subnet-address.

Parameter subnet-address is required, and for configuration simplicity, you configure the same subnet-address on both peer PE devices.

The commit script derives different IRB interface addresses per peer PE device based on subnet-address and peer-id, as follows:

  • peer-id 1: IRB interface address is subnet-address.

  • Otherwise, IRB interface address is subnet-address + 1 (add 1 to low-order address segment of the subnet address).

For example, if IPv4 subnet-address is 10.10.1.0/24:

  • On peer-id 1, address is 10.10.1.0/24

  • On peer-id 2, address is 10.10.1.1/24

Note:

The commit process doesn't do a commit check to enforce setting the same subnet address across both devices. By default, the commit script will derive the IRB interface subnet address based on the configured subnet-address at this hierarchy level.

As a result, we strongly recommend you use the same subnet address for this parameter on both peer PE devices, and only set the peer-id to be different on each device. This way the default address derivation works as expected for the IRB interfaces on each peer PE device.

IRB interface virtual gateway address

From set services evpn evpn-vxlan irb irb-instance subnet-address (inet | inet6) subnet-address

Include the following options at [edit services evpn evpn-vxlan irb irb-instance] to specify the virtual gateway IPv4 or IPv6 addresses instead of using the default derived virtual gateway addresses:

  • virtual-gateway-v4-address ipv4-virtual-gateway-address

  • virtual-gateway-v6-address ipv6-virtual-gateway-address

Derive the virtual gateway address for the IRB instance as the highest configurable address in the IPv4 or IPv6 subnet-address subnet range.

For example, if the IPv4 (subnet-address inet) subnet-address is 10.1.1.1/24, the derived virtual gateway IPv4 address is 10.1.1.254.

Similarly, for example, if the IPv6 (subnet-address inet6) subnet-address is 2001:db8::10:1:1:1/112, the derived virtual gateway IPv6 address is 2001:db8::10:1:1:fffe.

IRB interface virtual gateway MAC address

Include the following options at [edit services evpn global-parameters] to override the default virtual gateway IPv4 or IPv6 addresses:

  • virtual-gateway (v4-mac | v6-mac) virtual-gateway-mac-address—To set different virtual gateway MAC addresses for IPv4 traffic and IPv6 traffic

  • virtual-gateway-mac virtual-gateway-mac-address—To set same virtual gateway MAC address for both IPv4 traffic and IPv6 traffic

IPv4 virtual gateway MAC address: 00:00:5e:00:01:01

IPv6 virtual gateway MAC address: 00:00:5e:00:02:01

VXLAN network identifier (VNI) mapping

From:

  • set services evpn evpn-vxlan server server-name vlan-id-list [vlan-id ...]

  • set services evpn evpn-vxlan irb irb-instance vlan-id vlan-id

Add vlan-id to a base value of 10000

Underlay Peering for PE Devices—Default Protocol: EBGP

BGP group name: __SERVICES_EVPN_EVPN_VXLAN_EBGP_UNDERLAY

Autonomous system (AS) number

65000 + peer-id

Local AS number and peer AS number for BGP group configuration (local-as local-as, neighbor peer-as peer-as)

From set services evpn device-attribute peer-id peer-id.

On peer-id 1, local AS is 65000 + 4 and peer AS is AS is 65000 + 3.

The AS numbers are reversed on the other peer PE device peer-id 2: local AS is 65000 + 3 and peer AS is 65000 + 4.

Local device address and peer neighbor address for BGP group configuration (local-address, neighbor neighbor neighbor-address in generated configuration)

From set services evpn device-attribute peer-to-peer peer-subnet (inet | inet6) subnet-address.

On peer PE device peer-id 1, local address is peer-subnet subnet-address and neighbor address is subnet-address + 1.

The local address and neighbor address are reversed on the other peer PE device peer-id 2: local address is peer-subnet subnet-address + 1 and neighbor address is subnet-address.

MTU size for underlay or overlay peering interfaces

Use the overlay-mtu or underlay-mtu sizes you configure at the [edit services evpn global-parameters mtu] hierarchy level instead of using the default MTU size. The default MTU size varies based on platform and type of interface.

See Media MTU and Protocol MTU for more on MTU settings.

Routing policy to advertise loopback interface to EBGP peer PE device

 
set policy-options policy-statement EXPORT-LO0 term LOOPBACK from interface lo0.0
set policy-options policy-statement EXPORT-LO0 term LOOPBACK then accept
set policy-options policy-statement EXPORT-LO0 term REJECT then reject
set protocols bgp group __SERVICES_EVPN_EVPN_VXLAN_EBGP_UNDERLAY export EXPORT-LO

Overlay Peering for PE Devices—Default Protocol: EBGP

BGP group name: __SERVICES_EVPN_EVPN_VXLAN_EBGP_OVERLAY

Autonomous system (AS) number (for routing-options autonomous-system num)

From set services evpn device-attribute peer-id peer-id.

65000 + peer-id

Local device address and peer neighbor address for generated BGP group configuration (local-address local-address and neighbor neighbor-address)

From set services evpn device-attribute:

  • peer-id peer-id

  • loopback peer1-subnet peer1-subnet

  • loopback peer2-subnet peer2-subnet

Each peer device uses the provided loopback peerpeer-id-subnet addresses for the local address and its neighbor address.

Other Underlay Peering Options for PE Devices—OSPF

If you configure set services evpn device-attribute peer-to-peer underlay-connectivity ospf:

Aggregated Ethernet interface logical unit with OSPF underlay peering

ae0.0

OSPF area

0.0.0.0

Other Overlay Peering Options for PE Devices—IBGP

If you configure set services evpn device-attribute peer-to-peer overlay-connectivity ibgp:

Autonomous system number for internal BGP peering (set routing-options autonomous-system num in generated configuration

65000

Local device address and peer neighbor address for generated IBGP group configuration (local-address local-address and neighbor neighbor-address)

From set services evpn device-attribute:

  • peer-id peer-id

  • loopback peer1-subnet peer1-subnet

  • loopback peer2-subnet peer2-subnet

Same as with an EBGP overlay—each peer device uses the provided loopback peerpeer-id-subnet addresses for the local address and its neighbor address.

Option to Configure DHCP Relay for a VRF Routing Instance

If you configure set services evpn evpn-vxlan dhcp-relay name:

DHCP Relay group configuration

From set services evpn evpn-vxlan dhcp-relay name:

  • dhcp-server-address dhcp-server-address

  • relay-source relay-source-interface

  • vrf-instance instance-id

Derived DHCP relay group name is SERVICES_EVPN_EVPN_VXLAN_VRFname and default configuration is:

[edit routing-instances __SERVICES_EVPN_EVPN_VXLAN_VRF_instance-id forwarding-options]
set dhcp-relay forward-only
set dhcp-relay server-group SERVICES_EVPN_EVPN_VXLAN_VRFname dhcp-server-address

[edit routing-instances __SERVICES_EVPN_EVPN_VXLAN_VRF_instance-id forwarding-options dhcp-relay group SERVICES_EVPN_EVPN_VXLAN_VRFname]
set active-server-group SERVICES_EVPN_EVPN_VXLAN_VRFname
set overrides relay-source relay-source-interface
set relay-option-82 server-id-override

Easy EVPN LAG Configuration with Multihomed Servers

Figure 3 shows a topology with two multihomed servers connected to two peer PE devices. The peer PE devices are connected back-to-back in a small EVPN fabric that hosts two server VLANs (VLAN IDs 10 and 20).

Figure 3: Easy EVPN LAG Configuration with Two Multihomed Servers Easy EVPN LAG Configuration with Two Multihomed Servers

Simplified Configuration with Two Multihomed Servers

This section shows an example of a minimal easy EVPN LAG configuration for the topology in Figure 3. You use statements in the [edit services evpn] configuration stanza. This configuration provides all of the required parameters for the commit script to generate a corresponding EVPN-VXLAN configuration. You configure only a few elements that are specific to each peer PE device. Otherwise, most of the configuration is the same on both devices.

Note:

In this example, both peer PE devices host the same VLANs and use the same physical interface names to link to the multihomed servers, so those required configuration parameters can all be part of the common configuration across the peer PE devices. However, because physical interface allocations per device are usually different in actual customer deployments, we include the easy EVPN LAG configuration statements for those parameters separately below for each peer PE device.

The commit script uses default values for some elements. It also automatically derives other values for the generated configuration from the parameters in the simplified configuration, as mentioned earlier.

In this configuration, note that we configure the following common subnet address parameters across both peer PE devices. We do this because the commit script uses or automatically derives values on each peer PE device based on the peer-id on which the script is running:

  • The device loopback subnet addresses for each device:

    The commit script uses the provided loopback subnet address as the local address for the peer-id on which it runs. It uses the other peer-id loopback subnet address as its peer PE device neighbor address.

  • The peer PE device peer-to-peer link subnet addresses:

    The commit script uses the aggregated Ethernet interface subnet address for peer-id 1 as the configured subnet-address, and derives the address for peer-id 2 as that subnet-address plus 1 in the low-order byte of the address subnet range.

  • The IRB interface subnet address for each IRB interface that serves each VLAN:

    The commit script derives an IRB interface address in the same way as for the peer-to-peer aggregated Ethernet addresses above—it uses the configured subnet-address for peer-id 1, and adds 1 to the low-order byte of that subnet-address address range for peer-id 2.

Note:

The commit process doesn't do a commit check to enforce setting the same subnet address across both devices. By default, the commit script will automatically derive these addresses based on the provided subnet-address peer-id 1 values. As a result, we strongly recommend that you make sure to set these subnet addresses to the same base value on both peer PE devices in an easy EVPN LAG configuration. That way the default address derivation works as expected for the loopback device, loopback peer link, and IRB interface subnet addresses on each peer PE device.

See Derived Values in Generated Configuration for details on all of the values that the commit script derives for the generated configuration.

Easy EVPN LAG Configuration for Two Multihomed Servers

Peer PE 1:

Peer PE 2:

Common Configuration on Both Peer PE Devices:

Default Parameters Not Specified in the Simplified Configuration

The commit script uses the following default elements that you don't specify in this simplified configuration:

Table 3: Default Elements Used in Generated Configuration
Configuration Element Default Value

Aggregated Ethernet device count

255

Peer to peer PE device links

ae0

Overlay peering autonomous system number base value

65000

Underlay and overly peering protocol

EBGP

Underlay BGP group name

__SERVICES_EVPN_EVPN_VXLAN_EBGP_UNDERLAY

Underlay export policy name and policy statement

EXPORT-LO0

Overlay BGP group name

__SERVICES_EVPN_EVPN_VXLAN_EBGP_OVERLAY

EVPN-VXLAN MAC-VRF instance

instance-id 1

Name: __SERVICES_EVPN_EVPN_VXLAN_MAC_VRF_1

service-type vlan-aware

encapsulation vxlan

vtep-source-interface lo0.0

Virtual gateway MAC address (for IPv4)

00:00:5e:00:01:01

VLAN name

SERVICES_EVPN_EVPN_VXLAN_VLAN_vlan-id

Storm control profile name

__SERVICES_EVPN_EVPN_VXLAN_STORM_CONTROL

Generated Configuration with Two Multihomed Servers

With the easy EVPN LAG configuration commit script enabled, when you commit the configuration in Simplified Configuration with Two Multihomed Servers, the commit script generates the following EVPN-VXLAN configuration. By default, the commit script configures storm control and lightweight loop detection on the server-facing interfaces.

See Derived Values in Generated Configuration for all the values the commit script derives in this generated configuration from the simplified configuration.

Peer PE 1:

Peer PE 2:

Derived Values in Generated Configuration

In the generated configuration in Generated Configuration with Two Multihomed Servers, the commit script derives the following values based on corresponding easy EVPN LAG configuration elements:

Table 4: Derived Values in Generated Configuration Example with Two Multihomed Servers
Configuration Element Derived From [edit services evpn] Derived Value on Peer peer-id
Peer 1 Peer 2

AS number

Default AS number base: 65000

Underlay AS number: 65004

Overlay AS number: 65001

Underlay AS number: 65003

Overlay AS number: 65002

Peer to peer ae0 address

device-attribute peer-to-peer peer-subnet inet 10.1.1.0/31

10.1.1.0/31

10.1.1.1/31

Underlay EBGP local address and neighbor address

device-attribute peer-to-peer peer-subnet inet 10.1.1.0/31

local-address: 10.1.1.0/31

neighbor: 10.1.1.1/31

local-address: 10.1.1.1/31

neighbor: 10.1.1.0/31

Overlay EBGP local address and neighbor address

device-attribute loopback peer1-subnet 192.168.1.1/32 peer2-subnet 192.168.2.1/32

local-address: 192.168.1.1/32

neighbor: 192.168.2.1/32

local-address: 192.168.2.1/32

neighbor: 192.168.1.1/32

system-id for LACP configuration per server ESI LAG link

device-attribute system-id 10:11:12:13:14:10

ae1: 10:11:12:13:14:11

ae2: 10:11:12:13:14:12

ae1: 10:11:12:13:14:11

ae2: 10:11:12:13:14:12

Server VLANs for:

  • aex VLAN members

  • MAC-VRF instance VLAN members

  • IRB interface logical units

evpn-vxlan server SERVER_1 vlan-id-list [ 10 20 ]

evpn-vxlan server SERVER_2 vlan-id-list [ 10 20 ]

vlan-id 10

vlan-id 20

vlan-id 10

vlan-id 20

IRB interface subnet address and virtual-gateway-address (VLAN 10)

evpn-vxlan irb irb_10 subnet-address inet 10.10.1.1/24

10.10.1.1/24

10.10.1.254/24

10.10.1.2/24

10.10.1.254/24

IRB interface subnet address and virtual-gateway-address (VLAN 20)

evpn-vxlan irb irb_20 subnet-address inet 10.20.1.1/24

10.20.1.1/24

10.20.1.254/24

10.20.1.2/24

10.20.1.254/24

MAC-VRF instance route distinguisher

Default MAC-VRF instance-id: 1

device-attribute loopback peer1-subnet 192.168.1.1/32 peer2-subnet 192.168.2.1/32

192.168.1.1:1

192.168.2.1:1

MAC-VRF instance route target

Default MAC-VRF instance-id: 1

vrf-target:1:instance-id

vrf-target:1:1

vrf-target:1:1

MAC-VRF instance aggregated Ethernet interface logical units per server ESI LAG link:

ae[esi-lag-id].[instance-id]

Default MAC-VRF instance-id: 1

evpn-vxlan server SERVER_1 esi-lag-id esi-lag-id 1

evpn-vxlan server SERVER_2 esi-lag-id esi-lag-id 2

Server 1: ae1.1

Server 2: ae2.1

Server 1: ae1.1

Server 2: ae2.1

IRB interface names (VLANs 10 and 20)

evpn-vxlan irb irb_10 vlan-id 10

evpn-vxlan irb irb_20 vlan-id 20

evpn-vxlan server SERVER_1 vlan-id-list [ 10 20 ]

irb.10

irb.20

irb.10

irb.20

VLAN to VNI mappings (VLANs 10 and 20)

Default VNI base value: 10000

For VLAN 10: 10010

For VLAN 20: 10020

for VLAN 10: 10010

for VLAN 20: 10020

Add Configuration for a New Multihomed Server

Figure 4 shows the same topology as Figure 3 with an additional multihomed server.

The example configuration here shows how to add the new multihomed server, Server 3, that hosts the two VLANs VLAN 10 and VLAN 20.

Figure 4: Add a New Multihomed Server to an Existing Easy EVPN LAG Configuration Add a New Multihomed Server to an Existing Easy EVPN LAG Configuration

Simplified Configuration to Add a New Multihomed Server and ESI LAG

In this example, both peer PE devices:

  • Link to the new server using interface ge-0/0/5.

  • Host the same VLANs, VLAN 10 and VLAN 20.

  • Automatically derive the ES identifier by default (using the set interfaces aex esi auto-derive type-1-lacp command)

As a result, you can add the same additional easy EVPN LAG configuration statements on both devices, and the commit script generates the same additional configuration statements on both devices. You can also combine server server-name options at the [edit services evpn evpn-vxlan] hierarchy level into the same command, as this configuration example shows. You only need to add one configuration line item for this use case.

To add Server 3 with corresponding ESI LAG links, add the following single easy EVPN LAG configuration command to the existing simplified configuration on both Peer PE 1 and Peer PE 2:

Additional Generated Configuration for a New Multihomed Server and ESI LAG

The commit script generates the following additional configuration on both peer PE devices from the simplified configuration statements in Simplified Configuration to Add a New Multihomed Server and ESI LAG:

Add a New VLAN and IRB Interfaces

Figure 5 shows the same topology as Figure 5 with a new VLAN, VLAN 30, hosted only by Server 1 and Server 2.

Figure 5: Add a New VLAN and IRB Interfaces to an Existing Server Configuration Add a New VLAN and IRB Interfaces to an Existing Server Configuration

Simplified Configuration to Add a New VLAN

To add VLAN 30 for Server 1 and Server 2 to the original configuration in Easy EVPN LAG Configuration with Multihomed Servers), add the following easy EVPN LAG configuration commands:

On both Peer PE devices:

Note:

The simplified EVPN LAG configuration here is the same on both peer PE devices in this case because:

  • The commit script derives unique IRB interface subnet addresses for you using the provided subnet-address parameter.

  • We are adding the same VLAN on both devices.

Additional Generated Configuration for a New VLAN

The commit script generates the following additional configuration for VLAN 30:

Peer PE 1:

Peer PE 2:

Add a New Single-homed Server

Figure 6 adds a single-homed server, Server 4, that connects only to Peer PE 1. Server 4 hosts VLANs VLAN 10 and VLAN 20.

In the simplified configuration, you provide a single-home-id to identify single-homed servers links and the parameters related to that server. By default, the commit script uses aggregated Ethernet interface names starting with a base index of 1024 for links to single-homed servers. The commit script adds the single-home-id to that index, producing interface names that start with ae1025 (for single-home-id = 1).

Figure 6: Add Configuration for a Single-homed Server to an Existing Easy EVPN LAG Configuration Add Configuration for a Single-homed Server to an Existing Easy EVPN LAG Configuration

Simplified Configuration to Add a Single-homed Server

To add single-homed Server 4 to the original configuration in Easy EVPN LAG Configuration with Multihomed Servers), add the following easy EVPN LAG configuration commands on Peer PE 1:

Note:

You can combine server server-name options at the [edit services evpn evpn-vxlan] hierarchy level, as this configuration example shows. As a result, you only need to add one configuration line item for this use case.

Additional Generated Configuration for a Single-homed Server

The commit script generates the following additional configuration for Server 4 on Peer PE 1:

Use OSPF for the Underlay Configuration

By default, the easy EVPN LAG configuration commit script generates a configuration that uses EBGP for the underlay peering between the peer PE devices. See the __SERVICES_EVPN_EVPN_VXLAN_EBGP_UNDERLAY EBGP group configuration statements in the example in Generated Configuration with Two Multihomed Servers.

If you want to use OSPF for the underlay peering instead, include the following option in your easy EVPN LAG configuration on both peer PE devices:

With this option, in place of the default EBGP underlay configuration statements, the commit script generates an OSPF underlay peering configuration using the following default parameters (also refer to Table 2):

  • Aggregated Ethernet interface logical unit ae0.0

  • OSPF area 0.0.0.0

The commit script generates the following default OSPF underlay peering configuration on both peer PE devices:

Note:

If you don't want to use the default EBGP or OSPF underlay peering configurations, you can set the following easy EVPN LAG configuration option:

With this option set, the commit script will not generate any underlay peering configuration. In that case, you must manually configure the desired underlay peering.

Use IBGP for the Overlay Configuration

By default, the easy EVPN LAG configuration commit script generates a configuration that uses EBGP for the overlay peering between the peer PE devices. See the __SERVICES_EVPN_EVPN_VXLAN_EBGP_OVERLAY EBGP group configuration statements in the example in Generated Configuration with Two Multihomed Servers.

If you want to use IBGP for the overlay peering instead, include the following option in your easy EVPN LAG configuration on both peer PE devices:

With this option, instead of the default EBGP overlay configuration, the commit script generates an IBGP overlay peering configuration using the following default or derived parameters (also refer to Table 2):

  • IBGP group name __SERVICES_EVPN_EVPN_VXLAN_IBGP_OVERLAY

  • AS number 65000

  • IBGP local address and peer neighbor address derived from set services evpn device-attribute:

    • peer-id peer-id

    • loopback peer1-subnet peer1-subnet

    • loopback peer2-subnet peer2-subnet

For the example topology in Figure 3, the generated configuration is:

Peer PE 1:

Peer PE 2:

Related Documentation