BGP Path Selection, Route Reflectors and Policy

What it is

BGP chooses one best path for each prefix using attributes, and policy tools let you filter routes and change those attributes.

BGP does not use a simple metric. It compares a list of attributes in a fixed order, and the first difference decides. Policy is how an operator influences that order.

A route reflector (RR) is a router that is allowed to pass iBGP routes between iBGP peers, so a full mesh is not needed.

How it works

Think of best path selection as a sequence of tie breakers, and policy as a way to move the tie breaker you want.

The best path order

A path is considered only if its next hop is reachable. The router then compares paths in this order, on Cisco devices.

1. Highest weight, which is local to the router.
2. Highest local preference, which defaults to 100 and is shared in the AS.
3. Paths the router originated itself.
4. Shortest AS path.
5. Lowest origin code: IGP, then EGP, then incomplete.
6. Lowest MED, compared by default only between paths from the same neighbouring AS.
7. eBGP over iBGP.
8. Lowest IGP metric to the next hop.
9. Further tie breakers, ending with the lowest router ID and then the lowest neighbour address.

R1# show ip bgp
R1# show ip bgp 10.1.0.0/16

In the table, * means valid and > means best. A route with no * is usually invalid because the next hop is unreachable. The second command lists every path with its attributes and says why one was chosen.

Influencing path choice

R1(config)# route-map FROM-ISP-A permit 10
R1(config-route-map)# set local-preference 200
R1(config)# route-map TO-ISP-B permit 10
R1(config-route-map)# set as-path prepend 65002 65002
R1(config-route-map)# exit
R1(config)# router bgp 65002
R1(config-router)# neighbor 192.0.2.2 route-map FROM-ISP-A in
R1(config-router)# neighbor 198.51.100.2 route-map TO-ISP-B out
R1(config-router)# neighbor 192.0.2.2 weight 300

Local preference is only seen inside your AS, so it controls how traffic leaves. Prepending is how you discourage traffic from entering by one path.

Filtering

Filters apply per neighbour and direction. Three tools are common.

• A prefix list matches prefixes and lengths.
• A filter list matches the AS path with a regular expression.
• A route map matches many things and can also set attributes.

R1(config)# ip prefix-list NO-DEFAULT deny 0.0.0.0/0
R1(config)# ip prefix-list NO-DEFAULT permit 0.0.0.0/0 le 32
R1(config)# ip as-path access-list 10 permit ^65001$
R1(config-router)# neighbor 192.0.2.2 prefix-list NO-DEFAULT in
R1(config-router)# neighbor 192.0.2.2 filter-list 10 out

An AS path list uses ^ for the start and $ for the end. The pattern ^65001$ matches only routes whose whole path is that one AS.

To see what is sent or received, use the neighbour commands. Received routes before policy are only kept if soft-reconfiguration inbound is configured.

R1# show ip bgp neighbors 192.0.2.2 advertised-routes
R1# show ip bgp neighbors 192.0.2.2 received-routes
R1# show ip bgp regexp ^65001_

Communities tag routes so one policy can match many prefixes. They are not sent to a neighbour unless send-community is configured.

Route reflectors

Without an RR, an iBGP router will not pass on a route learned from an iBGP peer. The RR relaxes that rule for its clients.

RR(config-router)# neighbor 10.2.2.1 remote-as 65002
RR(config-router)# neighbor 10.2.2.1 route-reflector-client

The reflection rules are short:

• A route from a client goes to all clients and all non clients.
• A route from a non client goes to clients only.
• A route from an eBGP peer goes to all.

The RR adds two attributes to prevent loops: an originator ID and a cluster list. A router that sees its own ID in them drops the route.

The RR does not change the next hop, so clients must still reach it. This page covers one RR only.