BGP Multihoming ISP level redundancy

[jagdev.dhaliwal]

Hi All,

I want to go for ISP level redundancy,  with Dual internal Internet routers.  Public IP range is of /23. I want  BGP announcement of my discrete address block (e.g. /24) to their peers, in addition to the entire aggregated prefix (e.g. /23).

However I have only 2800 series routers with RAM of 768 MB (512+256).

Can you please suggest how should I proceed?

Thanks

Jagdev

[Raju Sekharan]

If you don't need to receive the complete BGP routes from the ISPs, then it is ok. You can request ISPs to advertise only defult-route to your BGP neighbors. Also make sure that you put a filter to accept only default-route from the ISPs

[jagdev.dhaliwal]

Please let know if the below config is good for my requirement

! Router 1:

!

autonomous-system 100

!

!

ip route 4.4.4.0  255.255.254.0 Null0 200

!

ip prefix-list  IP_OUT seq 10 permit 4.4.4.0/23

!

ip prefix-list DEFAULT_ROUTE_IN  seq 10 permit 0.0.0.0/0

!

router bgp 100

no synchronization

bgp log-neighbor-changes

bgp dampening

network 4.4.4.0  255.255.254.0

! define BGP session with ISP-1

neighbor 1.1.1.1 remote-as 200

neighbor 1.1.1.1 description BGP Transit to ISP-1

neighbor 1.1.1.1 timers 10 30

neighbor 1.1.1.1 version 4

neighbor 1.1.1.1 send-community

neighbor 1.1.1.1 soft-reconfiguration inbound

neighbor 1.1.1.1 prefix-list DEFAULT_ROUTE_IN in

neighbor 1.1.1.1 prefix-list IP_OUT out

neighbor 1.1.1.1 maximum-prefix 10

!

! define iBGP session

neighbor 4.4.4.2 remote-as 100

neighbor 4.4.4.2 description iBGP to Router2

neighbor 4.4.4.2 version 4

neighbor 4.4.4.2 send-community

neighbor 4.4.4.2 soft-reconfiguration inbound

neighbor 4.4.4.2 update-source Loopback0

!

end

++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

Router 2

!

autonomous-system 100

!

!

ip route 4.4.4.0  255.255.254.0 Null0 200

!

ip prefix-list IP_OUT seq 10 permit 4.4.4.0/23

!

ip prefix-list DEFAULT_ROUTE_IN  seq 10 permit 0.0.0.0/0

!

router bgp 100

no synchronization

bgp log-neighbor-changes

bgp dampening

network 4.4.4.0  255.255.254.0

! define BGP session with ISP-2

neighbor 2.2.2.1 remote-as 300

neighbor 2.2.2.1 description BGP Transit to ISP-2

neighbor 1.1.1.1 timers 10 30

neighbor 2.2.2.1 version 4

neighbor 2.2.2.1 send-community

neighbor 2.2.2.1 soft-reconfiguration inbound

neighbor 2.2.2.1 prefix-list DEFAULT_ROUTE_IN in

neighbor 2.2.2.1 prefix-list IP_OUT out

neighbor 2.2.2.1 maximum-prefix 10

!

! define iBGP session

neighbor 4.4.4.1 remote-as 100

neighbor 4.4.4.1 description iBGP to Router1

neighbor 4.4.4.1 version 4

neighbor 4.4.4.1 send-community

neighbor 4.4.4.1 soft-reconfiguration inbound

neighbor 4.4.4.1 update-source Loopback0

!

end

[Giuseppe Larosa]

Hello Jagdev,

your proposed configuration is fine, but you are only advertising the aggregate /23 and not the two component routes /24.

Also each router will advertise the /23 even if the link to the internal network is broken as you have a static route to null0. This provides stability of the prefix, but the price is that the aggregate is advertised even if the border router is isolated from the internal network. Having two border routers this can be less desirable.

In order to announce component routes and the aggregate you can do the following

router bgp 100

no network 4.4.4.0 255.255.254.0

network 4.4.4.0 255.255.255.0

network 4.4.5.0 255.255.255.0

aggregate-address 4.4.4.0 255.255.254.0

and you need to change the outbound prefix-list

ip prefix-list IP_OUT seq 10 permit 4.4.4.0/23 le 24

Note that if you want later to manipulate BGP path attributes (like doing selective AS path prepending) for each /24 component route it would be better to use a route-map in outbound instead of invoking the prefix-llst directly.

In this way all the changes can be done to the route-map and you can implement whatever policy you need.

Hope to help

Giuseppe

[jagdev.dhaliwal]

Thanks Giuseppe,

I want to publish one network network 4.4.4.0 255.255.255.0 would have primary link over ISP-1 (router1) and other network

4.4.5.0 255.255.255.0 would have primary link over ISP-2 (router2). In case one link goes down my netwrok should be reachable via other. What config changes i need to achive this?

Jagdev

[Giuseppe Larosa · Accepted solution]

Hello Jagdev,

You need to use a route-map applied outbound to eBGP neighbor to perform AS path prepending in a selective way

The route-map is ISP specific

border router 1:

ip prefix-list SECOND-ROUTE 4.4.5.0/24

route-map toISP1 permit 10

match ip address prefix  SECOND-ROUTE

set as-path prepend 100 100 100

route-map toISP1 permit 20

match ip address prefix IP_OUT

router bgp 100

no neigh 1.1.1.1 prefix-list IP_OUT out

neigh 1.1.1.1 route-map toISP2 out

In the same way for border router 2:

ip prefix-list FIRST-ROUTE 4.4.4.0/24

route-map toISP2 permit 10

match ip address prefix FIRST-ROUTE

set as-path prepend 100 100 100

route-map toISP2 permit 20

match ip address prefix IP_OUT

router bgp 100

no neigh 2.2.2.2 prefix-list IP_OUT out

neigh 2.2.2.2 route-map toISP2 out

Hope to help

Giuseppe