ECMP Support for L3VPN
Overview
Equal Cost Multipath (ECMP) support for L3VPN helps in load-balancing the traffic for L3VPN routes while managing traffic distribution across multiple paths of equal cost. It enables improved load balancing and resiliency across VPN traffic paths.
Currently, in Inter-AS VPN configuration options A, B, and C, the best path is programmed in fast path for L3VPN routes. Using ECMP, multiple paths are configured in fast path while also balancing the traffic load for a given L3VPN route.
Feature Characteristics
The PE routers PE1, PE2, ... P8 learn IPv4 and IPv6 routes connected to VRF on the User Network Interface. The protocol BGP running in the devices publish these routes to UPE as VPNv4/VPNv6 routes using Multi Protocol BGP with Route Distinguisher (RD) and Route Target (RT). The user facing PE installs the next-hops to its Routing Information Base (RIB) and Forwarding Information Base (FIB) and prepares the hashing keys based on the L3 fields of the incoming traffic, and also load balances the traffic at service level and transport (LDP/RSVP/SR) level.
Figure 34. L3VPN ECMP Characteristics
The Hardware Service Layer (HSL) programs the MPLS forwarding entities in hardware, maintains the software linking of the programmed entities and is responsible for load balancing of L3VPN traffic. It resolves Forward Error Correction (FEC) and uses the hierarchical FEC resolution along with the ECMP group table. The existing L3 keys provided by OcNOS are used to load balance the traffic, and MPLS Label stack is used to load balance the traffic at transit nodes.
Load balancing is done at service and transport level and multiple paths are maintained at L3VPN for the given nexthops. The first load balance is done at service level hierarchy and the second load balance is done at transport level hierarchy, such as LDP, RSVP, and SR.
The existing load balance keys are used at the PE ingress nodes and the incoming label stack is used at the P transit nodes (ASBR and RR) to load balance the traffic.
Supported Hardware
Qumran1, Qumran2, and J2C+ devices.
Limitations
-
Load balancing of transit routers (RR and ASBRs) in case of inter-AS is not supported.
-
Multipath EVPN Integrated Routing and Bridging (IRB) is not supported.
Prerequisites
1. | Define Interfaces and Loopback Addresses |
Configure Layer 2 interfaces such as port channel interfaces, for example po1, and assign specific IP addresses for proper identification and routing. Additionally, assign loopback IP addresses to establish essential points of connectivity.
!
interface lo
ip address 127.0.0.1/8
ip address 8.8.8.8/32 secondary
ipv6 address ::1/128
interface po7
ip address 31.1.1.8/24
2. | Configure IGP for Dynamic Routing |
Enable ISIS to facilitate dynamic routing on all nodes within the network. Define ISIS router instances to match loopback IP addresses and add network segments to ISIS areas for proper route distribution. Set up neighbor relationships using loopback IP addresses, ensuring efficient route advertisement and convergence for optimal network performance.
ISIS Configuration
router isis 1
is-type level-2-only
metric-style wide
microloop-avoidance level-2
mpls traffic-eng router-id 8.8.8.8
mpls traffic-eng level-2
capability cspf
dynamic-hostname
bfd all-interfaces
net 49.0000.0000.0008.00
passive-interface lo
!
interface po7
mpls ldp-igp sync isis level-2
isis network point-to-point
ip router isis 1
OSPF Configuration
router ospf 1
ospf router-id 8.8.8.8
network 8.8.8.8/32 area 0.0.0.0
network 31.1.1.0/24 area 0.0.0.0!
!
interface po7
ip ospf network point-to-point