Cisco Support Community
cancel
Showing results for 
Search instead for 
Did you mean: 
Announcements

Welcome to Cisco Support Community. We would love to have your feedback.

For an introduction to the new site, click here. If you'd prefer to explore, try our test area to get started. And see here for current known issues.

New Member

address-family ipv4 in bgp/eigrp/rip

Hi,

I would like to know how eigrp and rip used address-family. My understand of the utilisation of address-family in BGP is that the routing control plane used BGP communitiies to indicate traffic classes and update associated topologies. I think that this means the BGP carries the RD or RT (do not know which one) to identify to which VRF belongs the announce route. Please correct me if I am wrong.

I would like to understand if eigrp or RIP carries similar information with their route of VRF is only provides local significance on the router.

Finally, I would like to know if the address-family concept applies to OSPF.

Thanks for your help

Stephane

  • MPLS
5 REPLIES
Hall of Fame Super Silver

Re: address-family ipv4 in bgp/eigrp/rip

Hello Stephane,

a BGP VPNv4 prefix is made by RD:IPv4 with extended community RT (at least one but multiple are possible)

so both are present with RD being part of 96 bit VPNv4 address

EIGRP and RIP uses address-family in order to provide support for VRF PE-CE protocol: the idea is  that of creating a small sub database that is VRF specific. The routing messages sent out and received on PE-CE link are regular IPv4 updates and only PE node is aware of the VRF existance.

In order to make the routes known in the backbone (export phase) you need to redistribute the PE-CE IGP into BGP in the appropriate address family (address family ipv4 vrf )

Without doing this the VPNv4 routes are not created and the routes learned by CE are not available on remote PE nodes.

In the same way to pass remote VRF sites routes to the local CE using EIGRP/RIP/OSPF you need to redistribute BGP ( implies same address family ipv4 vrf into the PE-CE IGP routing protocol or if the design allows this, to generate a default route.

First implementantions of OSPF as PE-CE required a dedicated OSPF process for each VRF meaning a clear scalabilty limit.

http://www.cisco.com/en/US/docs/ios/mpls/configuration/guide/mp_cfg_layer3_vpn_ps6441_TSD_Products_Configuration_Guide_Chapter.html#wp1062224

later implementations have been able to avoid to count these processes under the box limit but I think they still use process + vrf name in process creation they just remove that limit.

the reason for the missing of address family in OSPF comes from its specifications that provide closed data structures, it is the same reason a new version of OSPF ( OSPFv3) has been introduced for IPv6

RIP and EIGRP probably allows for an internal tag to be added to type length value data structures so the capability to introduce address families.

Hope to help

Giuseppe

New Member

Re: address-family ipv4 in bgp/eigrp/rip

Hi Giuseppe,

Thanks for the answer.

Can you just please describe the difference between a bgp session using vpn4 and ipv4 as in the following configuration.

Conf address vpn4 + address-family ipv4                      Conf  address-family ipv4 only

router bgp 100                                                           router bgp 800
bgp log−neighbor−changes                                         address-family ipv4 vrf v12
neighbor 10.10.10.6 remote−as 100                             neighbor 83.0.0.0 remote as-100
neighbor 10.10.10.6 update−source Loopback0             neighbor 83.0.0.3 activate
!                                                                                network 8.8.2.0 mask 255.255.255.0
address−family vpnv4                                                  redistribute ospf 2 match internal
neighbor 10.10.10.6 activate
neighbor 10.10.10.6 send−community both
exit−address−family
!
address−family ipv4 vrf Customer_B
redistribute connected
no auto−summary
no synchronization
exit−address−family

Can you please tell me if BGP running only ipv4 only is like RIP/EIGRP where the vrf is only used locally to separate route belonging to different VRF.Does these BGP packets carries something different than regular BGP.

I have taken the address-family ipv4  configuration from the following link:

http://www.cisco.com/en/US/docs/switches/lan/catalyst4500/12.2/20ew/configuration/guide/vrf.html.

BGP is used to transport the OSPF customer VRF route from CE-1 (AS-800) to PE (AS-100). The example does not show how router are transport from PE to CE-2. I assume it is done by running another BGP session from PE (AS-100) to CE-2 (AS-400) but I am not sure. If this configuration is like any standard BGP with a local VRF context, I would assume that BGP could been replaced by EIGRP or RIP.

Thanks

Stéphane

Re: address-family ipv4 in bgp/eigrp/rip

Stephane,

The fact is that you cant replace BGP with any other routing protocol because its a Multi Protocol BGP. The Multi Protocol BGP allows to carry Labeled IPv4 packets , Labeled IPv6 Packets, VRF, Multicast and Unicast. The Labeled Packets cant be carried out by other routing protocol, Its Only BGP have this capability.

The RT (Route-target) which is extended BGP community attribute is carried out through VPNv4 or vpnv6 sessions and it cant be also carried out by another routing protocol.

HTH

Mohamed

New Member

Re: address-family ipv4 in bgp/eigrp/rip

Hi Guys,

If I have correctly understand these answers. The BGP define under address-family ipv4 uses Network Layer Reachability (NLRI) composed of a standard IPv4 address associated with it`s mask. It can exchange route with another BGP peer on a different router that is not defined under an  address-family ipv4.

BGP define under address-family vpn4 are using a Network Layer Reachability (NLRI) composed of VPNv4 prefix(RD +IPv4) address,Route Target that carried by the Extended Communities (enabled with neighbor X.X.X.X send community both) and MPLS mux label.

In addition (this is the part where I am not sure),  every routes that belongs to the routing procol defined under the address-familly will be redistributed to the neighbor defined under address-family vpn using VPNv4 prefix.


Thanks for your help

Stephane

Cisco Employee

Re: address-family ipv4 in bgp/eigrp/rip

Hello Stephane,

If I have 
correctly understand these answers. The BGP define under address-family 
ipv4 uses Network Layer Reachability (NLRI) composed of a standard IPv4 
address associated with it`s mask. It can exchange route with another 
BGP peer on a different router that is not defined under an  
address-family ipv4.

Correct. BGP defined within address-family ipv4 is the usual BGP. In fact, this way of configuring BGP is called the AFI model (Address Family Identifier), see this document:

http://www.cisco.com/en/US/docs/ios/12_2t/12_2t15/feature/guide/ftbhycli.html

Having some configuration placed in address-family ipv4 is just having a nice "box" for the configuration that was usually placed at the global level of BGP configuration without any address families.

BGP define under 
address-family vpn4 are using a Network Layer Reachability (NLRI) 
composed of VPNv4 prefix(RD +IPv4) address,Route Target that carried by 
the Extended Communities (enabled with neighbor X.X.X.X send community 
both) and MPLS mux label.

Correct. As a side remark, note that the BGP is actually the only protocol that runs over and above VRFs. All IGPs run confined inside VRFs and do not see beyond the VRF. The BGP is the only protocol that keeps the overview of all VRFs at once as it is able to preserve their isolation when carrying the routes from the various VRFs.

In addition (this is the part where I am not sure),  every routes that 
belongs to the routing procol defined under the address-familly will be 
redistributed to the neighbor defined under address-family vpn using 
VPNv4 prefix.

Correct. The redistribution from a particular VRF into BGP will be configured in the BGP's address-family ipv4 vrf NAME configuration section and the redistributed routes will be subsequently advertised to neighbors activated under the address-family vpnv4 context.

Best regards,

Peter

2791
Views
13
Helpful
5
Replies
This widget could not be displayed.