Добавил:
Опубликованный материал нарушает ваши авторские права? Сообщите нам.
Вуз: Предмет: Файл:
Internet Routing Architectures Second Edition - Cisco press.pdf
Скачиваний:
102
Добавлен:
24.05.2014
Размер:
4.91 Mб
Скачать

Internet Routing Architectures, Second Edition

Route Reflector

Route reflector configuration may now be specified on a per-AF basis, providing a considerable amount of flexibility over the old "centralized" method. This section discusses both old and new styles of configuring route reflection.

Old Style

With the old style, route reflector (RR) client properties were specified globally, and the configuration applied to all AFs negotiated with its clients. The RR knew that it had to reflect routes to and from clients by specifying route-reflector-client for a particular neighbor or IBGP peer group. The following is an example in which the IBGP peer 1.1.1.1 is made a route reflector client for both unicast and multicast IPv4 prefixes:

router bgp 109

neighbor 1.1.1.1 remote-as 109 nlri unicast multicast neighbor 1.1.1.1 route-reflector-client

AF Style

The AF style of configuring router reflects the fact that a neighbor or peer group is a route reflector client is AF-dependent and is configured in AF mode. In other words, just because a peer is a route reflector client in IPv4 unicast mode does not automatically make it an RR client in IPv4 multicast mode. It must now be specified by explicitly configuring the client in IPv4 multicast AF mode. Thus, the preceding configuration that made 1.1.1.1 a client for both unicast and multicast would now be expressed as follows:

router bgp 109

neighbor 1.1.1.1 remote-as 109 neighbor 1.1.1.1 route-reflector-client

!

address-family ipv4 multicast neighbor 1.1.1.1 activate

neighbor 1.1.1.1 route-reflector-client

The new mode gives the operator the flexibility to make a router the route reflector for only certain AFs. As such, the route reflector topologies for different AFs can vary.

Aggregation

Configuring aggregation for single or multiple AFs was complex if varying policies were required. This section provides information on both old and new methods of configuring aggregation.

page 407

Internet Routing Architectures, Second Edition

Old Style

In the old style, a multicast aggregate was configured the same way you configured a unicast aggregate—via the aggregate-address command. The aggregate-address command was enhanced to specify whether the aggregate address should be applied to unicast or multicast using the nlri keyword in the aggregate-address command. The following is an example of generating an aggregate in the multicast BGP table:

router bgp 109

aggregate-address 174.0.0.0 255.0.0.0 as-set nlri multicast

The nlri options that can be specified on an aggregate-address command are unicast, multicast, and unicast multicast, which generate aggregates in the BGP table, in the multicast BGP table, or in both, respectively. The absence of the nlri keyword in the aggregate-address command results in the aggregate's being generated as unicast only.

AF Style

The presence of AF modes for different AFs eliminates the need for the nlri keyword in the aggregate-address command. The AF mode under which the aggregate is specified determines the table where the aggregated prefix should be generated. Hence, the aggregate displayed in the old-style section can be generated in the AF style as follows:

router bgp 109

!

address-family ipv4 multicast aggregate-address 174.0.0.0 255.0.0.0 as-set

List of BGP Commands

Table D-1 lists the BGP commands and the category of each command/subcommand.

Table D-1. BGP Commands/Subcommands Grouped by Category

Command/Subcommand

Category

address-family ipv4 unicast

Can appear once under router bgp mode

address-family ipv4 multicast

Can appear once under router bgp mode

aggregate-address

Per AF

auto-summary

Per AF

bgp always-compare-med

Global to BGP

bestpath

Global to BGP

client-to-client

Per AF

cluster-id

Global to BGP

confederation

Global to BGP

dampening

Per AF

default

Global to BGP

deterministic-med

Global to BGP

page 408

Internet Routing Architectures, Second Edition

fast-external-fallover

Global to BGP

log-neighbor-changes

Global to BGP

redistribute-internal

Global to BGP

router-id

Global to BGP

scan-time

Per AF

default-metric

Global to BGP

distance

Per AF

maximum-paths

Per AF

neighbor

Applied on a per-neighbor basis

activate

Per AF

advertisement-interval

Global to the neighbor (session)

default-originate

Per AF (policy)

description

Global to the neighbor

distribute-list

Per AF (policy)

ebgp-multihop

Global to the neighbor (session)

filter-list

Per AF (policy)

local-as

Global to the neighbor (session)

maximum-prefix

Per AF (policy)

next-hop-self

Global to the neighbor (session)

password

Global to the neighbor (session)

peer-group

Per AF (policy)

prefix-list

Per AF (policy)

remote-as

Global to the neighbor (session)

remove-private-AS

Global to the neighbor (session)

route-map

Per AF (policy)

route-reflector-client

Per AF

send-community

Per AF (policy)

shutdown

Global to the neighbor (session)

soft-reconfiguration

Per AF (policy)

timers

Global to the neighbor (session)

update-source

Global to the neighbor (session)

version

Global to the neighbor (session)

weight

Per AF (policy)

Network

Per AF

Redistribute

Per AF

Synchronization

Per AF

table-map

Per AF

Timers

Global to BGP

Upgrading to the AF Style

In order to have a smooth upgrade path, support has been added to parse the old-style 12.0S commands (which had the nlri keyword). These commands are as follows:

neighbor

network

aggregate

set nlri and match nlri in route maps

page 409

Internet Routing Architectures, Second Edition

The only caveat is that the old-style commands can be used as long as no new features need to be activated. In that event, the old-style BGP commands need to be translated to the new style.

To migrate to the new command set, you must enter the bgp upgrade-cli command in router configuration mode:

Router(config-router)#bgp upgrade-cli

This command will translate the old configuration to the new one. As always, a wr mem will need to be done in order to save the new configuration. (Note: The bgp upgrade-cli command is not shown in the configuration.)

page 410