Feature Requirements |
Series |
Models |
|---|---|---|
In an EVPN over VXLAN IPv4/IPv6 dual-stack scenario, if the number of received Add-Path routes with the same prefix exceeds the upper limit and the number of advertised Add-Path routes reaches the upper limit, the RR prefers only the maximum number of Add-Path routes allowed and advertises these routes to IPv4 or IPv6 peers. The remaining routes are not advertised. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
You are advised to configure an IRT in a maximum of 128 L3VPN, L3EVPN, or MVPN instances. If it is configured in more than 128 L3VPN, L3EVPN, or MVPN instances, excessive route leaking may occur, causing memory insufficiency. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
In EVPN E-LAN scenarios, MPLS EVPN VPLS and SRv6 cannot co-exist on an instance. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
Option B is not supported in the EVPN E-LAN over SRv6 PE scenario. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
In EVPN E-LAN over SRv6 BE scenarios, remote FRR in single-active dual-homing mode is not supported. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
In EVPN E-LAN over SRv6 BE scenarios, only the ACL filtering based on the IPv6 and public network addresses is supported on the network side. ACL filtering based on private network addresses is not supported. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
EVPN L3VPNv4 routes can recurse to either SRv6 BE or MPLS tunnels, instead of both SRv6 BE and MPLS tunnels. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
EVPN L3VPNv6 routes can recurse to either SRv6 BE or MPLS tunnels, instead of both SRv6 BE and MPLS tunnels. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
In an L3 EVPN scenario, if an IP route with a prefix specified using the gateway-ip field recurses to multiple MPLS tunnels, only one recursive tunnel is used to forward traffic. It is not recommended that an IP route with a prefix specified using the gateway-ip field recurse to multiple MPLS tunnels working in load balancing mode. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
In an EVPN+VLL scenario, if a VE interface is configured on an SPE, the main interface must be bound to the VLL and the sub-interface must be bound to the EVPN. If the type of the VLL AC interface on a UPE and that of the EVPN AC interface on the SPE are both QinQ in asymmetrical mode, EVPN traffic cannot be forwarded between the UPE and SPE. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
In the scenario of interworking between VLL and EVPN in non-decoupling mode, a VLL supports only the following tunnel types: LDP, TE, and LDP over TE. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
The EVPN public network interface does not support MF classification. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
EVPN loop protection depends on MAC address learning. If MAC address learning is disabled, the function becomes invalid. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
In EVPN and VPLS interworking scenarios, MAC FRR is not supported on the PW side. If MAC FRR is enabled based on EVPN, PW-side MAC addresses point to the PW interface, instead of the protection link. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
IGMP snooping is not supported in the EVPN and VPLS interworking scenario. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
In EVPN and VPLS interworking scenarios, only the primary/secondary PW+EVPN single-active mode is supported. EVPN active-active is not supported. The PW standalone mode is also not supported. The secondary PW cannot be configured with the ignore-standby command. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
In a scenario where VPLS interworks with an EVPN in full-mesh networking, a loop occurs in either of the following scenarios: (1) VPLS is configured to work in spoken mode. (2) VPLS E-Trunks are configured to work in active-active mode. Configure VPLS PWs to work in hub mode and VPLS E-Trunks to work in master/backup mode. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
The PBB-EVPN does not support public network BGP tunnels. Change the tunnel policy, so that PBB-EVPN routes are not iterated to BGP tunnels. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
The PBB-EVPN network does not support the differentiated service mode. Set priorities for the upstream and downstream nodes. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
When a VLL is connected to an EVPN, the loopback scheme is used in the downstream direction. In this case, CAR rate limit is inaccurate on a PW-VE sub-interface in the downstream direction. Please run the "evpn access vll convergence separate disable" command. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
When a VSI is bound to multiple BDs, if an EVPN instance is bound to one of the BDs, the VSI's PW is set to Down. As a result, services in the other BDs that are not bound to EVPN instances are interrupted. When a VSI is bound to multiple BDs, unbind the BDs to be bound to EVPN instances from the VSI. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
In an active-active scenario where primary and secondary PWs of a VPLS are connected to a VXLAN, the "ignore-standby-state" parameter cannot be configured for the PW on the device where the VPLS is connected to the VXLAN. If both the parameter and dual-receive are configured, duplicate multicast traffic is generated in the direction from the VXLAN to the VPLS. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
If the PE at one end of a VPLS has only one VSI and the PE at the other end has multiple VSIs, the VPLS network cannot be reconstructed into an EVPN. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
For a BD bound to both an EVPN instance and a VPLS VSI, the VPLS PWs cannot be set to be in Spoken mode. If a VPLS PW is set to be in Spoken mode, loops may occur. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
The EVPN switch to which a user device is connected is replaced with a VPLS network switch. Before the MAC entry aging time expires, the user device is reconnected to the EVPN switch. As a result, services are interrupted until the VPLS network ages the MAC entry for the user device. When the VPLS network is reconstructed into the EVPN, manually shut down the network device interface connecting to the switch and then re-enable the interface after the network reconstruction. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
An EVPN instance and a VSI are bound to the same BD. If the BD receives a MAC Withdraw packet for a PW from the remote VSI, the BD deletes the MAC address learned from the AC interface but does not clear the MAC address learned from the EVPN instance. Deleting the MAC address learned from the AC interface triggers MAC address deletion on the EVPN. After the BD relearns a MAC address from the AC interface, the BD broadcasts the MAC address on the EVPN. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
In an EVPN dual-homing scenario, using dynamic ESIs cannot ensure fast fault-triggered switching, and switching depends only on protocol convergence. Therefore, using static ESIs is recommended. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
In an EVPN active-active scenario where a CE is dual-homed to PEs, if a BGP connection between the dual-homing PEs goes down, the dual-homing PEs both enter the DF state. As a result, the BUM traffic doubles, and the dual-homing and active-active mechanism fails. Do not reset BGP on dual-homing PEs. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
If an EVPN instance is bound to a BD that has been bound to a VSI for an existing VPLS service, services may be interrupted. After the EVPN instance is bound to the BD, services are restored. Manually set the VPLS PW to Down. In master/backup E-Trunk scenarios, bind an EVPN instance to a BD on the backup E-Trunk first and then perform the binding on the master E-Trunk. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
For the EVPN E-LAN service, the control word can only be statically configured. The same control word must be configured on devices. For the EVPN VPWS service, a device checks whether the control word carried in a received route matches the local control word. A control word inconsistency causes a failure to create an EVPN VPWS service. Control word self-negotiation (configured using the "preferred control-word" command) is not supported. In EVPN VPWS dual-homing networking, a bypass path needs to be established. Therefore, the control word configuration must be consistent on all involved PEs (when a bypass path is established between a single-homing PE and each of dual-homing PEs and between dual-homing PEs). A control word inconsistency causes a failure to establish bypass paths. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
In the L2 EVPN access to L3 EVPN scenario, a created L3VE main interface inherits the system MAC address by default. To connect to L2VE interfaces mapped to different L3VE main interfaces in a BD, an interface MAC address must be separately configured for each L3VE main interface. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
When routes over which VLL VPN QoS traffic travels recurse to BGP LSPs, the BGP entropy label function is not supported. In the basic VPLS forwarding, VPLS VPN QoS, or PBB VPLS scenario or when BGP LSPs function as public network tunnels for VPLS, the BGP entropy label function is not supported. When the decoupling between the public network and private network is not enabled for both VPLS and MPLS in the scenario of VPLS routes recursive to BGP LSPs, the BGP entropy label function is not supported for known unicast traffic. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
When an EVPN E-LAN SRv6 egress is dual-homed to two PEs, the egress protection function does not take effect for BUM traffic. Therefore, locators need to be planned so that the egress protection function takes effect only for known unicast routes. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
1. Egress protection is not supported in the CE multi-homing scenario. 2. Different locators need to be deployed for single-homing and dual-homing services. Egress protection is not required for the locators used by single-homing services. 3. The BUM service of the EVPN E-LAN service and the unicast service use different locators, and the locator used by the BUM service is not configured with egress protection. 4. In the scenario where one route is leaked to multiple VPN instances, the remote SID can be mapped only to one local service instance. Ensure that the route leaking results on the dual-homing PEs are the same. (The RT used to locally leaked routes between VPN instances on a dual-homing PE must be the same as that on the other dual-homing PE. The local route leaking relationship configured on a dual-homing PE must be the same as that on the other dual-homing PE.) The egress protection function may fail. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
In an EVPN dual-homed single-active DF scenario, CFM needs to be deployed between the CE and PE to improve the switchback performance. The CE and dual-homed PEs need to support the association between DF election with CFM to clear MAC addresses on the CE. Otherwise, the switchback performance cannot be ensured. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
PBB EVPN public network tunnels do not support the following scenarios: 1. TE over GRE 2. LDP over TE over GRE |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
In a sampling scenario where SRv6 and MPLS interworks in Option B mode: 1. In the SRv6 -> MPLS direction, the inbound interface restored through outbound interface sampling is the master interface, and the outbound interface restored through inbound interface sampling is the master interface. 2. In the MPLS -> SRv6 direction, the inbound interface restored through outbound interface sampling is the master interface. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
EVPN in BD VLAN aware mode does not support SRv6. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
When an ESI has multiple sub-interfaces and AC status to influence DF election is enabled, the non-revertive function for ESI priority election does not take effect. BUM traffic on some sub-interfaces may not meet the expectation. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
In a BD EVPN scenario, IGMP snooping does not support synchronization of user leave events between dual-homing PEs on the access side through EVPN Leave Synch routes (BGP EVPN Type 8 routes). Per-flow Eth-Trunk deployment is recommended on the downstream CE to prevent join and leave messages from being sent to different devices between the dual-homing PEs on the access side. If the IGMP Report message sent by the CE arrives at one PE and the Leave message sent by the CE arrives at the other PE, the PEs cannot respond to the Leave messages immediately, which delays the leave process. Instead, the PEs can only wait for the expiry of the user entries. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
Regarding EVPN active-active: 1. Dual-homing devices do not support inter-AS interconnection. 2. Non-anycast VXLAN active-active is not supported. 3. The ESIs of different interfaces on the dual-homing side must be different. 4. In active-active scenarios, a CE must be directly connected to PEs, and no transmission device can be deployed between the CE and PEs. Traffic switching performance in case of a link failure is affected. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
The E-Tree attributes configured on the two dual-homing access interfaces must be the same. Otherwise, traffic may flow back. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
When a VLL accesses an EVPN, only non-BD EVPN (MPLS EVPN) is supported. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
Regarding EVPN E-Line: The EVPN single-active mode blocks traffic on the AC interface. If the E-Trunk master/backup status is determined, the interface is set to Down. Therefore, EVPN single-active access and other active-active or Layer 3 services cannot use the same link for dual-homing access. If an EVPN works in single-active mode, the EVPN single-active services and active-active or Layer 3 services cannot be transmitted through the same link. If they use the same link, traffic for other services may not be forwarded properly. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
In the EVPL smooth evolution scenario, detection using ping/tracert evpn vpws -r 4 is not supported. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
1. Egress protection does not support CE multi-homing. 2. Different locators need to be deployed for single-homing and dual-homing services. Egress protection is not deployed for the locator used by single-homing services. 3. EVPN E-LAN BUM and unicast services use different locators, and egress protection is not deployed for the locator used by BUM services. 4. When a route is leaked into multiple VPN instances, the remote SID can be mapped to only one local service instance. Ensure that the route leaking results on the dual-homing PEs are the same. (The RTs or route import relationships between instances on the dual-homing PEs must be the same.) Egress protection may fail. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
In an IBGP scenario, if segmented VXLAN is configured on PE1, PE1 reflects the inclusive route learned by the local CE to PE2 through an RR. As a result, an extra VXLAN tunnel destined for the CE is generated on PE2. To solve this problem, you need to configure a routing policy on PE2 to deny the inclusive route on the CE side. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
In a distributed gateway deployment scenario where both Layer 2 and Layer 3 services are transmitted (a BDIF interface is bound to an EVPN L3VPNv6 Layer 3 instance and a BD is bound to an EVPN Layer 2 instance) and HEVPN is configured, local traffic cannot be forwarded along the shortest path. Basic protocol exchange and traffic forwarding are not affected in the scenario where both Layer 2 and Layer 3 services are transmitted and HEVPNv6 is configured. Only local forwarding along the shortest path is affected. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
In a BGP EVPN L3VPN over SR-MPLS BE scenario, primary and backup SR-MPLS BE tunnels exist between an EVPN L3VPN edge PE and a peer edge PE. If fast tunnel fault detection is not enabled and the peer edge PE fails, protocol hard convergence triggers a primary/backup tunnel switchover. If multiple types of tunnels (for example, LDP LSPs) exist, you need to configure a tunnel policy to speed up route convergence. The tunnel policy focuses only on the status of the primary and backup SR-MPLS BE tunnels. You need to configure the following tunnel policy in the VPN view. When the primary SR-MPLS BE tunnel fails, traffic is quickly switched to the backup SR-MPLS BE tunnel, and other types of tunnels (such as LDP tunnels) are not selected. The following configurations are recommended: tunnel-policy AAA Configure a tunnel policy on the tunnel select-seq sr-lsp ldp load-balance-number 1 // to select only SR-MPLS BE tunnels. # ip vpn-instance vpn-801 ipv4-family route-distinguisher 100:801 tnl-policy AAA ip frr vpn-target 801:801 export-extcommunity evpn vpn-target 801:801 export-extcommunity vpn-target 801:801 import-extcommunity evpn vpn-target 801:801 import-extcommunity tnl-policy AAA evpn evpn mpls routing-enable In an EVPN L3VPN over SR-MPLS BE scenario where fast tunnel switching detection (for example, BFD detection) is not enabled, an edge PE of an EVPN-L3VPN restarts and depends on protocol hard convergence. During the restart, traffic is switched to the backup SR-MPLS BE tunnel, and this process is slow. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
SRv6-enabled VRFs cannot coexist with Layer 3 EVPN in VXLAN mode. Interworking between a VXLAN data center network (VXLAN L3 EVPN) and a DCI backbone network (SRv6 L3 EVPN) is not supported in DCI scenarios. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
EVPN SRv6 instances do not support EVPN IGMP, which is mutually exclusive with the igmp-snooping proxy configuration. EVPN instances enabled with SRv6 do not support EVPN IGMP. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
In an EVPN L3VPN scenario, if an IP route with a prefix specified using the gateway-ip field recurses to multiple MPLS tunnels, only one recursive tunnel is used to forward traffic. In an EVPN L3VPN scenario, when a GWIP is used for route recursion, you are not advised to select a GWIP that will be recursed to multiple MPLS tunnels for load balancing. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
EVPN peers and instances do not support outbound policy-based routing in mpls-label mode. Configure the inbound route-policy in mpls-label mode for the EVPN instance and EVPN neighbor that receives EVPN routes. The EVPN route-policy fails to uniformly configure the outbound route-policy in mpls-label mode for all the remote devices. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
EVPN L3VPNv6 does not support the import of VPN labeled IPv6 routes. Level-2 carriers' hierarchical access to EVPN L3VPNv6 is not supported. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
Two EVPN instances with the same ERT are configured on a PE. The PE advertises EVPN routes to its EVPN peers. The EVPN peers select an optimal route and send BUM traffic to the local PE. In this case, traffic may fail to be forwarded. Configure different ERTs for different EVPN instances on a PE. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
EVPN SRv6 instances do not support the mLDP PMSI, which is mutually exclusive with the inclusive-provider-tunnel configuration. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
In a BGP EVPN distributed VXLAN gateway scenario, the gateways at both ends of a VXLAN tunnel cannot be configured with the same static ARP entry or the same static route with the next hop being iterated to the VXLAN tunnel. Otherwise, a loop may occur. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
A PW-VE interface cannot be bound to an EVPL instance in SRv6 mode. EVPN VPWS over SRv6 cannot interwork with traditional VPWS. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
The cloud VPN and DCI scenarios support co-existence of different scenarios. ARP/IRB routes: When vBDIF interfaces are configured on a cloud VPN in an L2+L3 scenario, the following situations may occur. A route-policy needs to be configured to prevent routes from being re-originated. 1. Layer 2 services belong to different EVPN instances. Layer 3 services belong to the same EVPN instance but are on different network segments. In this case, a route filter policy can be configured for the network segments. 2. Layer 2 services belong to different EVPN instances, and Layer 3 services belong to different VPN instances. In this case, a route filter policy must be configured for the routes originated from the DC side. IP prefix routes: Configure a route-policy to prevent routes from being re-originated. For details, see the description of situations 1 and 2 for ARP/IRB routes. The re-originated routes on the DCI network are advertised to the cloud VPN, which affects service traffic planning. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
The evpn-vpws ignore-ac-state command used to ignore the AC interface status is in a dual-homing network deployed with 1:1 protection. This command does not apply to dual-homing access on one side and single-homing access on the other side. In case of dual-homing access on one side and single-homing access on the other side, if an AC-interface goes down on the dual-homing side, routes are not withdrawn. If bypass attributes are not configured, packet loss occurs. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
VPLS supports LDPs. EVPN supports only MPLS EVPN. VPLS and EVPN can be integrated using the same BD. After PWs access EVPN, the learned MAC addresses can be sent to the control plane. MAC FRR and IRB routes are not supported. Redirection based on the MAC addresses learned through PWs is not supported. Use commands to allow BDs to be bound to the VPLS only with the type of MPLS EVPN or LDP. Use commands to disable the system from supporting IRBs. VXLAN EVPN and VPLS cannot be bound to the same BD. IRB routes cannot be learned. Redirection based on the MAC addresses learned by PWs is not supported. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
In a scenario where routes are imported between the VPN and public network instances, the color attribute cannot be changed based on the route-policy. The imported routes inherit the route recursion result of the source VPN instance. Specifically, the source route is iterated based on the SR-MPLS TE Policy even after import-rib is configured. If a configuration is not supported by the source route, it is also not supported by the imported routes. Import routes based on a specified protocol type or network segment to prevent route import between the VPN and public network instances. The coloring correctness in a scenario where routes are imported between the VPN and public network instances is affected. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
EVPN does not support the following route re-origination scenarios: Common L3VPN over SRv6 and EVPN L3VPN over SRv6 EVPN L3VPN over MPLS and EVPN L3VPN over MPLS EVPN VPLS over MPLS and EVPN VPLS over SRv6 EVPN over VXLAN and EVPN over SRv6 |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
Instance-based outbound policy-based routing does not apply to ES and ES-AD routes. Configure coloring by specifying the route type (ES or ES-AD routes) based on peer+. The correctness of ES and ES-AD route recursion to SR-policy tunnels is affected. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
Except for SPEs in HEVPN scenarios, labels are distributed by EVPN instance, instead of by route or next hop. Basic protocol interaction and traffic forwarding are not affected. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
In an EVPN dual-homing scenario, using dynamic ESIs cannot ensure fast fault-triggered switching, and switching depends only on protocol convergence. Therefore, using static ESIs is recommended. Configuring dynamic ESIs cannot ensure high switching performance in an EVPN dual-homing scenario. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
An ESI is applied to multiple sub-interfaces on multi-homing PEs connected to a CE, and the AC status is configured to influence DF election. In this situation, the non-revertive mode for ESI preference-based DF election may fail on some sub-interfaces, and BUM traffic on these sub-interfaces is switched back after the primary DF recovers. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
In the E2E SRv6 telecom cloud scenario, if FRR is enabled on PEs and the cost values of Add-Path routes advertised by different leaf nodes are different, the PEs generate merely ECMP paths, but not FRR+ECMP paths. Traffic may be interrupted. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
In the L3VPN CSC scenario, the level-1 carrier PE needs to support per-route per-label, but the EVPN L3VPN does not support per-route per-label. Therefore, the EVPN L3VPN CSC scenario is not supported. Do not deploy EVPN L3VPN CSC on PEs. In an EVPN L3VPN CSC scenario, traffic fails to be forwarded. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
VLL access to SRv6 BD EVPN VPLS is not supported. After the segment-routing ipv6 best-effort command is run for an EVPN instance in BD mode and the instance is bound to a BD, the BD cannot be bound to any PW VE interface or sub-interface. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
In the E2E SRv6 telecom cloud scenario, if FRR is enabled on PEs and the cost values of Add-Path routes advertised by different leaf nodes are different, the PEs generate merely ECMP paths, but not FRR+ECMP paths. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
EVPN in a non-BD mode does not support EVPN VPLS over SRv6. SRv6 cannot be enabled for EVPN instances in a non-BD mode, that is, the segment-routing ipv6 best-effort command cannot be run for the instances. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
When a BD VPLS network is dual-homed to an EVPN through PWs, you are not advised to configure an ESI in the dual-homing access-side PW view. Do not configure an ESI in the dual-homing access-side PW view. In this scenario, if the same ESI is configured in the access-side PW view, the switchback performance is poor when the master device is faulty, and traffic is interrupted. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
According to the standard protocol, EVPN routes carrying the GW-IP attribute cannot carry labels or SIDs. Therefore, the following scenarios are not supported when the GW-IP attribute is specified for EVPN routes using the route export policy: PE1 reflects the route to PE2 and PE3 through the same RR. PE2 is expected to receive the route and iterate the route to the private IP address. PE3 is expected to receive the route and iterate the route to the SR tunnel. Routes are reflected by different RRs, and routing policies are specified for multiple neighbors. Traffic cannot be forwarded along the path that is expected to be iterated to an SR tunnel. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
Interworking between VXLAN/VXLAN EVPN and SRv6 BD EVPN VPLS is not supported. After the segment-routing ipv6 best-effort command is run in a BD-EVPN instance and the instance is bound to a BD, the BD cannot be bound to a VNI. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
When VXLAN6 and VXLAN4 coexist, if the router-id-extend private enable command needs to be run on the local end, ensure that the remote end can preferentially select IPv4 unicast MAC routes. Implement proper planning. When VXLAN6 and VXLAN4 coexist, if the router-id-extend private enable command is run and IPv6 unicast MAC routes are preferentially selected, remote unicast traffic is interrupted. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
For EVPN VPLS services, only static control words can be configured. The configurations must be consistent between devices. The EVPN VPWS service allows routes to carry control word information for negotiation. If the control words do not match, the service cannot be established. This service does not support control word auto-negotiation (preferred control-word). In an EVPN VPWS dual-homing scenario, a bypass path generally needs to be established. The control words must be the same between the single-homing PE and dual-homing PEs and between dual-homing PEs. Otherwise, the bypass path cannot be established. Configure the control word according to the preceding restrictions. If the control word is not configured accordingly, services may fail. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
In the E2E SRv6 telecom cloud scenario, if FRR is enabled on PEs and the cost values of Add-Path routes advertised by different leaf nodes are different, the PEs generate merely ECMP paths, but not FRR+ECMP paths. Traffic loops may occur. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
In the scenario where L3 EVPN and L3VPN coexist, a device by default sends the routes of an L3VPN instance to both the BGP-EVPN address family and the BGP-VPNv4/v6 address family (used in local route leaking). In the scenario where L3 EVPN and L3VPN coexist, to enable a device to send the routes of an L3VPN instance only to the BGP-EVPN address family but not to the BGP-VPNv4/v6 address family, you need to configure an export route-policy in the L3VPN instance IPv4/v6 address family. By default, routes of an L3VPN instance are sent to the BGP-VPNv4 address family only. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
In the EVPN E-Tree dual-homing scenario, the leaf attributes of AC interfaces bound to an EVPN instance must be the same; otherwise, a traffic loop occurs, and the device reports the EVPN_ETREE_MULTIHOMED_CONFIG_ERROR event. Configure the same leaf attribute for such AC interfaces. The traffic forwarding in the EVPN E-Tree scenario is adversely affected. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
ESI preference-based DF election is not supported in the VPWS EVPN scenario. The VPWS EVPN scenario is not supported. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
In the E2E SRv6 telecom cloud scenario, if FRR is enabled on PEs and the cost values of Add-Path routes advertised by different leaf nodes are different, the PEs generate merely ECMP paths, but not FRR+ECMP paths. FRR protection and ECMP cannot coexist. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
EVPN VPLS over SRv6 TE is not supported. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
EVPN instances in BD VLAN-aware mode do not support SRv6. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
1. Only EVPN E-LAN and E-Line E-Trunk single-active scenarios are supported. The EVPN instance deployed in E-LAN mode is in BD EVPN mode. Other non-BD modes are not supported. 2. Only EVC sub-interfaces can be used for access, and E-Trunk single-active scenarios are deployed. Other VPLS or VLL single-active scenarios are not supported. The handover performance cannot be guaranteed, and many packets are lost during the handover. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
In the E2E SRv6 telecom cloud scenario, if FRR is enabled on PEs and the cost values of Add-Path routes advertised by different leaf nodes are different, the PEs generate merely ECMP paths, but not FRR+ECMP paths. FRR protection and ECMP cannot coexist. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
The non-revertive mode for ESI preference-based DF election must be enabled in the same ESI view on multi-homing PEs. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
Interworking between traditional VPLS and SRv6 BD EVPN VPLS is not supported. After a BD-EVPN instance is bound to a BD using the segment-routing ipv6 best-effort command, the BD cannot be bound to a VSI. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
The Option B scenario is not supported. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
MPLS VPLS EVPN Option B and SRv6 are mutually exclusive. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
In the scenario where an Eth-Trunk is dual-homed to an EVPN network in single-active mode, LACP is deployed on the CE side. It is recommended that different ESIs be configured on the dual-homing access side. In this scenario, if the same ESI is configured on the access side, the switchback performance of the master device is low. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
After the mac-learn-style qualify command is run for a BD, the VLAN encapsulation type and traffic action on a sub-interface bound to the BD are constrained as follows: 1, If the encapsulation type is dot1q, the pop outer action is not supported. 2, If the encapsulation type is dot1q, the pop outer action is not supported. 3, If the encapsulation type is QinQ, the pop double action is not supported. 4, If the encapsulation type is untag, none of the flow action is supported. 5, If the encapsulation type is any other, none of the flow action is supported. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
In an EVPN L3VPNv6 scenario, IPv6 prefix routes cannot carry the gateway IP address as a link-local address. In the typical DCGW-TOR-VNF networking in symmetric forwarding scenarios of the telco cloud, if IPv6 services are used, routes cannot be advertised between VNFs and TOR switches using IS-IS IPv6 or OSPFv3. If IPv6 services are used, static routes must be configured between VNFs and TOR switches. In the symmetric forwarding scenario of the telco cloud, when IPv6 services are used, dynamic paths cannot be deployed between VNFs and TOR switches. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
A maximum of 1024 EVPL instances can be associated with the same VRRP group. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
ESI and DF election result-based master/backup determination are mutually exclusive on the slot VE interface: 1, If an ESI has been configured for the slot VE interface, DF election result-based master/backup determination cannot be configured for the static ESI. 2, If DF election result-based master/backup determination has been configured for a static ESI, the ESI cannot be configured on the slot VE main interface. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
Co-existence of the EVPN and VSI: 1, In the co-existence scenario, only dual-homing single-active EVPN access is supported. 2, In the co-existence scenario, only independent PW redundancy is supported on the network side. 3, In the co-existence scenario, MAC mobility data update may be delayed. 4, In the co-existence scenario, MAC Withdraw signaling cannot be initiated when an AC interface goes down. Packet loss takes a long time in switching scenarios, and traffic switching is performed only after MAC address entries age. 5, In the co-existence scenario, VPLS multicast cannot co-exist with the EVPN. 6, In the co-existence scenario, the MPLS LSR ID and EVPN source address must be the same. 7, A VSI can co-exist only with EVPN over MPLS/EVPN over SRv6. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
VXLAN cannot interwork with EVPN over SRv6. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
For L2VE interfaces, EVPN VPLS does not support dual-homing networking: 1, When an L2VE sub-interface is bound to a BD, no ESI can be configured for the L2VE main interface. 2, When an ESI is configured for an L2VE main interface, the L2VE sub-interface cannot be bound to a BD. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
If the mac-learn-style qualify command is run in a BD, the BD cannot be bound to a VNI. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
The mac-learn-style qualify command configuration in a BD is mutually exclusive with the VBDIF interface configuration. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
After the mac-learn-style qualify command is run in a BD, the VSI bound to the BD cannot have the peer upe command configuration. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
The bypass-vxlan enable command cannot be configured for the EVPN instance on the stitching node. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
In DCI scenarios, EVPN ping supports only segment-by-segment ping, and does not support end-to-end ping or EVPN tracert. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
In an EVPN dual-homing single-active scenario where E-Trunk is used to determine the active/standby roles of devices and dual-homing PEs use dynamic ESIs, the max active-linknumber 1 command cannot be run on the CE.Downlink traffic may be interrupted. In an EVPN dual-homing single-active scenario where E-Trunk is used to determine the active/standby roles of devices and dual-homing PEs use dynamic ESIs, the max active-linknumber 1 command cannot be run on the CE. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
EVPN ring network access is not supported in non-BD EVPN, and EVPN VPWS scenarios. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
EVPN ring network access is supported only by dot1q EVC sub-interfaces. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
Interworking between EVPN and VSI: 1. In interworking scenarios, BGP AD VPLS and BGP VPLS cannot access EVPN, and only LDP VPLS can access EVPN. 2. In interworking protection scenarios, only PW redundancy in master/slave mode can be used for EVPN access, and PW redundancy in independent mode cannot. 3. VPLS in qualify mode and EVPN cannot interwork. 4. In interworking scenarios, the SPE does not support VBDIF interfaces. 5. In interworking scenarios, only Y.1731, Y.1564, and RFC 2544 E2E detection capabilities are supported. Ping and tracert are not supported. 6. Only EVPN over MPLS/EVPN over SRv6 can interwork with VSI. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
VXLANv6 cannot interwork with EVPN over SRv6. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
In an EVPN VPWS scenario, different EVPN instances are bound to different EVPL instances. If the same local service ID and remote service ID are configured for the EVPL instances, the same RT cannot be configured for different EVPN instances. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
The mac-learn-style qualify command and the arp collect host enable/nd collect host enable/arp l2-proxy enable command are mutually exclusive in a BD. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
The mac-learn-style qualify command configuration in a BD is mutually exclusive with the BD tag configuration in an EVPN instance bound to the BD. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
EVPN E-LAN/E-Line services are dual-homed to DFs in single-active mode to determine the primary/secondary status. 1. Aggregation and mixed non-direct connection scenarios are not supported. 2. A CE cannot be dual-homed to PEs through an Eth-Trunk in DF dual-homing and single-active scenarios. 3. When PWs access an EVPN or an EVPN interworks with a VPLS, the DF single-active scenario is not supported. 4. The EVPN standard DF and traditional VPLS cannot coexist on the network. 5. VXLAN EVPN does not support the single-active mode. Therefore, the single-active mode of DFs is not supported. 6. Only dual-homing single-active DFs are supported, and multi-homing single-active DFs are not supported. 7. The DF cannot be switched to the active state and the CE cannot clear MAC address entries. To communicate with non-Huawei devices, the CE needs to age MAC address entries or learn reverse flows. 8. After a DF switchover is performed, CFM instructs the CE to clear Layer 2 MAC addresses, but the CE cannot send gratuitous ARP packets to instruct the CE to learn new ARP entries. Therefore, the CE cannot be dual-homed to the PEs through VBDIF interfaces. 9. Only BD EVPN dual-homing in single-active mode is supported, and the primary and secondary DFs can be specified. 10. Dynamic ESI dual-homing in single-active mode is not supported. 11. In global single-active mode, the DF cannot be specified to determine the active/standby status. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
VRRP can be configured only for EVPL instances whose AC interfaces are of the slot VE type: 1, If a non-slot VE sub-interface is bound to an EVPL instance, VRRP cannot be configured for the EVPL instance. 2, If VRRP has been configured for an EVPL instance, non-slot VE sub-interfaces cannot be bound to the EVPL instance. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
A PW-VE interface cannot be bound to an EVPL instance in SRv6 mode. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |