As is well known, in an MPLS VPN network, all PE routers only maintain IPv4 routing information required by directly connected VPNs. This characteristic is achieved by importing only specific VPN-IPv4 routes, which are related to the RT configuration in VRFs. However, this implementation creates a problem: the MP-BGP routing table on a PE router maintains all BGP routes, even those not used by any VRF. This is a significant waste of resources, both in terms of PE router memory and in advertising this information across the backbone network to PE routers that do not use it.
Based on this understanding, MP-BGP uses several mechanisms to avoid the problem described above: “Automatic Route Filtering (ARF) on PE Routers”: Since PE routers may receive routing information they do not need, we want to be able to filter incoming MP-IBGP routing updates at the ingress. The Automatic Route Filtering feature meets this filtering need. When the RT value contained in a received MP-IBGP route does not match any VRF configured on the PE, the PE router discards it to reduce the amount of information stored in memory. This feature is enabled by default on all PE routers and requires no additional configuration to activate.
If the PE router also acts as a route reflector for other PE routers, or if the PE router simultaneously serves as an MP-EBGP peer that needs to exchange inter-domain VPN information, we need to disable this feature using the command: no bgp default route-filter ” BGP Route Refresh: Because the Automatic Route Filtering feature is enabled on PE routers, when the PE router’s configuration changes鈥攕uch as adding a new VRF or deleting an existing VRF鈥攖he PE router’s “demand” for VPN routes may not be met in a timely manner. Therefore, BGP provides a route refresh mechanism to achieve this goal. When this feature is used, a short time after the configuration change, the PE router will request its MP-IBGP neighbors to resend routing updates to obtain the missing VPN-IPv4 information. At the same time, we can also use this function by manually refreshing VPN routing information with the command clear ip bgp * vpnv4 unicast in. After entering this command, the router will send a route refresh request to all PE neighbors.
ORF: The Route Refresh and Automatic Route Filtering (ARF) features can reduce the amount of routing information PE routers need to maintain, but useless routing information still propagates through the network and is only discarded upon reaching specific PE routers. To avoid wasting network resources, BGP provides the ORF feature to advertise its inbound route filters to neighbors, so that unnecessary routing information can be filtered out before reaching the receiving PE router. The ORF field is exchanged through OPEN messages during session establishment between two PE routers.
Separating MP-IBGP Sessions Between PE Routers Each PE router only needs to receive routes for the VPNs they serve. Therefore, we can divide PE routers into independent MP-IBGP clusters within the MPLS VPN network based on different VPN service objects, with PE routers in the same cluster establishing full-mesh MP-IBGP sessions.
Route Reflectors and Confederations In large networks, full-mesh BGP sessions impose significant limitations on network scalability. Route reflectors and confederations can be used to reduce the number of MP-IBGP sessions within a domain.
Route Reflector Groups: By configuring an RT list on the route reflector to specify particular VPN routes, and by configuring route reflector groups, the route reflector utilizes ORF to advertise its inbound route filters, thereby preventing PE client routers from sending useless routes to the route reflector. Related command: bgp rr-group {extcom-list-number}
銆怑ditor’s Picks銆?/strong>