I am hoping to get some direction on a routing design.
I have a data center connected to two collocations via DWDM. The three locations are within in the same EIGRP AS 10, EIGRP is the only routing protocol running between the data center and the two collocations. Each collocation is running BGP with it's own AS (ie collo A is BGP 1 and collo B is BGP 2) and also EIGRP 10. Each collo has a Verizon MPLS circuit and a XO MPLS circuit, the MPLS routers and core at each collo have iBGP and EIGRP running between them.
Collo A is the primary collo for the collo A networks, data center networks, outbound internet and a backup for collo B networks. It is also the primary transit point between the Verizon and XO MPLS networks (XO is used as a backup for critical offices)
Collo B is the primary collo for the collo B networks and backup for the data center networks, outbound internet and collo A networks. It is also the backup transit point between the Verizon and XO MPLS networks.
Here is where I am running into an issue. I will use collo B as the example.
I want to group the traffic in my outbound bgp route map separately so that I can take different actions on each type of traffic (collo A, collo B, data center, etc) and change it at any point with minimal configuration for any given traffic type.
In collo B I want to prepend collo A networks and data center networks but I do not want to use a prefix-list, especially for the data center because that would become too cumbersome with the amount of networks I would need in the list.
My original plan was to tag collo A eigrp routes with a tag of 1, tag data center internal eigrp routes with 10 and data center external eigrp routes with 20. Then on the collo B core redistribute eigrp into bgp and match internal and tag 20. Then on the routers have a route map statement for data center traffic matching tag 10 and 20 and prepend the routes, then do the same with a route-map statement for the collo A routes.
However matching a tag on an outbound BGP map is not supported. So I am looking for another way to accomplish this. My thought was to use a community and on the routers have an inbound BGP map on the core switch peer matching the tags and setting a BGP community and then on the outbound map match that community and set the prepend.
router bgp 9100
bgp log-neighbor-changes
neighbor 10.2.0.238 remote-as 9100
neighbor 10.2.0.254 remote-as 9100
neighbor 1.1.1.1 remote-as 65000
neighbor 1.1.1.1 fall-over bfd
!
address-family ipv4
aggregate-address 10.2.0.0 255.255.128.0 summary-only
neighbor 10.2.0.238 activate
neighbor 10.2.0.238 next-hop-self
neighbor 10.2.0.238 soft-reconfiguration inbound
neighbor 10.2.0.238 route-map SET_COM in
neighbor 10.2.0.254 activate
neighbor 10.2.0.254 next-hop-self
neighbor 10.2.0.254 soft-reconfiguration inbound
neighbor 10.2.0.254 route-map SET_COM in
neighbor 1.1.1.1 activate
neighbor 1.1.1.1 soft-reconfiguration inbound
neighbor 1.1.1.1 route-map BGP_IN in
neighbor 1.1.1.1 route-map BGP_OUT out
exit-address-family
!
route-map SET_COM permit 5
description *** Set Community ***
match tag 10
set community 65000:3
!
route-map SET_COM permit 10
description *** Set Community ***
match tag 20
set community 65000:3
!
route-map BGP_OUT permit 50
description *** Data Center Routes to Allow and Actions ***
match community 65000:3
set as-prepend 9100 9100 9100
Is this a workable solution? Can the community be anything as long as it is not something that matches a community within the ISP? Attached is a diagram as well.
Kevin
