Configuring Multiprotocol BGP (MP-BGP) Support for CLNS

By default, commands entered under the router bgp command apply to the IPv4 address family. This will continue to be the case unless you enter the no bgp default ipv4-unicast command as the first command under the router bgp command. The no bgp default ipv4-unicast command is configured on the router to disable the default behavior of the BGP routing process exchanging IPv4 addressing information with BGP neighbor routers.

The configuration of MP-BGP support for CLNS allows BGP to be used as an interdomain routing protocol in networks that use CLNS as the network-layer protocol. This feature was developed to solve a scaling issue with a data communications network (DCN) where large numbers of network elements are managed remotely. For details about the DCN issues and how to implement this feature in a DCN topology, see the DCN Network Topology.

BGP, as an Exterior Gateway Protocol, was designed to handle the volume of routing information generated by the Internet. Network administrators can control the BGP routing information because BGP neighbor relationships (peering) are manually configured and routing updates use incremental broadcasts. Some interior routing protocols such as Intermediate System-to-Intermediate System (IS-IS), in contrast, use a form of automatic neighbor discovery and broadcast updates at regular intervals.

CLNS uses network service access point (NSAP) addresses to identify all its network elements. Using the BGP address-family support, NSAP address prefixes can be transported using BGP. In CLNS, BGP prefixes are inserted into the CLNS Level 2 prefix table. This functionality allows BGP to be used as an interdomain routing protocol between separate CLNS routing domains.

Implementing BGP in routers at the edge of each internal network means that the existing interior protocols need not be changed, minimizing disruption in the network.

The figure below shows a generic BGP CLNS network containing nine routers that are grouped into four different autonomous systems (in BGP terminology) or routing domains (in OSI terminology). To avoid confusion, we will use the BGP terminology of autonomous systems because each autonomous system is numbered and therefore more easily identified in the diagram and in the configuration discussion.

Figure 1

Components in a Generic BGP CLNS Network

Within each autonomous system, IS-IS is used as the intradomain routing protocol. Between autonomous systems, BGP and its multiprotocol extensions are used as the interdomain routing protocol. Each router is running either a BGP or Level 2 IS-IS routing process. To facilitate this feature, the BGP routers are also running a Level 2 IS-IS process. Although the links are not shown in the figure, each Level 2 IS-IS router is connected to multiple Level 1 IS-IS routers that are, in turn, connected to multiple CLNS networks.

Each autonomous system in this example is configured to demonstrate various BGP features and how these features work with CLNS to provide a scalable interdomain routing solution. In the figure above, the autonomous system AS65101 has a single Level 2 IS-IS router, R1, and is connected to just one other autonomous system, AS65202. Connectivity to the rest of the network is provided by R2, and a default route is generated for R1 to send to R2 all packets with destination NSAP addresses outside of AS65101.

In AS65202 there are two routers, R2 and R3, both with different external BGP (eBGP) neighbors. Routers R2 and R3 are configured to run internal BGP (iBGP) over the internal connection between them.

AS65303 shows how the use of BGP peer groups and route reflection can minimize the need for TCP connections between routers. Fewer connections between routers simplifies the network design and the amount of traffic in the network.

AS65404 shows how to use redistribution to communicate network reachability information to a Level 2 IS-IS router that is not running BGP.

The configuration tasks and examples are based on the generic network design shown in the figure above. Configurations for all the routers in the figure are listed in the Implementing MP-BGP Support for CLNS Example.

The Multiprotocol BGP (MP-BGP) Support for CLNS feature can benefit a DCN managing a large number of remote SONET rings. SONET is typically used by telecommunications companies to send data over fiber-optic networks.

The figure below shows some components of a DCN network. To be consistent with the BGP terminology, the figure contains labels to indicate three autonomous systems instead of routing domains. The network elements--designated by NE in Figure 2--of a SONET ring are managed by OSI protocols such as File Transfer, Access, and Management (FTAM) and Common Management Information Protocol (CMIP). FTAM and CMIP run over the CLNS network-layer protocol, which means that the routers providing connectivity must run an OSI routing protocol.

Figure 2

Components in a DCN Network

IS-IS is a link-state protocol used in this example to route CLNS. Each routing node (networking device) is called an intermediate system (IS). The network is divided into areas defined as a collection of routing nodes. Routing within an area is referred to as Level 1 routing. Routing between areas involves Level 2 routing. Routers that link a Level 1 area with a Level 2 area are defined as Level 1-2 routers. A network element that connects to the Level 2 routers that provide a path to the DCN core is represented by a gateway network element--GNE in Figure 2. The network topology here is a point-to-point link between each network element router. In this example, a Level 1 IS-IS router is called an NE router.

Smaller Cisco routers such as the Cisco 2600 series were selected to run as the Level 1-2 routers because shelf space in the central office (CO) of a service provider is very expensive. A Cisco 2600 series router has limited processing power if it is acting as the Level 1 router for four or five different Level 1 areas. The number of Level 1 areas under this configuration is limited to about 200. The entire Level 2 network is also limited by the speed of the slowest Level 2 router.

To provide connectivity between NE routers, in-band signaling is used. The in-band signaling is carried in the SONET/Synchronous Digital Hierarchy (SDH) frame on the data communications channel (DCC). The DCC is a 192-KB channel, which is a very limited amount of bandwidth for the management traffic. Due to the limited signaling bandwidth between network elements and the limited amount of processing power and memory in the NE routers running IS-IS, each area is restricted to a maximum number of 30 to 40 routers. On average, each SONET ring consists of 10 to 15 network elements.

With a maximum of 200 areas containing 10 to 15 network elements per area, the total number of network element routers in a single autonomous system must be fewer than 3000. Service providers are looking to implement over 10,000 network elements as their networks grow, but the potential number of network elements in an area is limited. The current solution is to break down the DCN into a number of smaller autonomous systems and connect them using static routes or ISO Interior Gateway Routing Protocol (IGRP). ISO IGRP is a proprietary protocol that can limit future equipment implementation options. Static routing does not scale because the growth in the network can exceed the ability of a network administrator to maintain the static routes. BGP has been shown to scale to over 100,000 routes.

To implement the Multiprotocol BGP (MP-BGP) Support for CLNS feature in this example, configure BGP to run on each router in the DCN core network--AS64800 in Figure 2--to exchange routing information between all the autonomous systems. In the autonomous systems AS64600 and AS64700, only the Level 2 routers will run BGP. BGP uses TCP to communicate with BGP-speaking neighbor routers, which means that both an IP-addressed network and an NSAP-addressed network must be configured to cover all the Level 2 IS-IS routers in the autonomous systems AS64600 and AS64700 and all the routers in the DCN core network.

Assuming that each autonomous system--for example, AS64600 and AS64700 in Figure 2--remains the same size with up to 3000 nodes, we can demonstrate how large DCN networks can be supported with this feature. Each autonomous system advertises one address prefix to the core autonomous system. Each address prefix can have two paths associated with it to provide redundancy because there are two links between each autonomous system and the core autonomous system. BGP has been shown to support 100,000 routes, so the core autonomous system can support many other directly linked autonomous systems because each autonomous system generates only a few routes. We can assume that the core autonomous system can support about 2000 directly linked autonomous systems. With the hub-and-spoke design where each autonomous system is directly linked to the core autonomous system, and not acting as a transit autonomous system, the core autonomous system can generate a default route to each linked autonomous system. Using the default routes, the Level 2 routers in the linked autonomous systems process only a small amount of additional routing information. Multiplying the 2000 linked autonomous systems by the 3000 nodes within each autonomous system could allow up to 6 million network elements.

The Multiprotocol BGP (MP-BGP) Support for CLNS feature adds the ability to interconnect separate OSI routing domains without merging the routing domains, which provides the capability to build very large OSI networks. The benefits of using this feature are not confined to DCN networks, and can be implemented to help scale any network using OSI routing protocols with CLNS.

1.    enable

2.    configure terminal

3.    router bgp as-number

4.    no bgp default ipv4-unicast

5.    neighbor {ip-address | peer-group-name} remote-as as-number

6.    address-family nsap [unicast]

7.    neighbor ip-address activate

8.    end

1.    enable

2.    configure terminal

3.    router isis area-tag

4.    net network-entity-title

5.    is-type [level-1 | level-1-2 | level-2-only]

6.    end

When a router running IS-IS is directly connected to an eBGP neighbor, the interface between the two eBGP neighbors is activated using the clns enable command, which allows CLNS packets to be forwarded across the interface. The clns enable command activates the End System-to-Intermediate System (ES-IS) protocol to search for neighboring OSI systems.

Note	Running IS-IS across the same interface that is connected to an eBGP neighbor can lead to undesirable results if the two OSI routing domains merge into a single domain.

When a neighboring OSI system is found, BGP checks that it is also an eBGP neighbor configured for the NSAP address family. If both the preceding conditions are met, BGP creates a special BGP neighbor route in the CLNS Level 2 prefix routing table. The special BGP neighbor route is automatically redistributed in to the Level 2 routing updates so that all other Level 2 IS-IS routers in the local OSI routing domain know how to reach this eBGP neighbor.

To configure interfaces that are being used to connect with eBGP neighbors, perform the steps in this procedure. These interfaces will normally be directly connected to their eBGP neighbor.

1.    enable

2.    configure terminal

3.    interface type number

4.    ip address ip-address mask

5.    clns enable

6.    no shutdown

7.    end

1.    enable

2.    configure terminal

3.    interface type number

4.    ip address ip-address mask

5.    clns router isis area-tag

6.    ip router isis area-tag

7.    no shutdown

8.    end

1.    enable

2.    configure terminal

3.    router bgp as-number

4.    no bgp default ipv4-unicast

5.    neighbor {ip-address | peer-group-name} remote-as as-number

6.    address-family nsap [unicast]

7.    network nsap-prefix [route-map map-tag]

8.    neighbor ip-address activate

9.    end

1.    enable

2.    configure terminal

3.    router isis area-tag

4.    net network-entity-title

5.    redistribute protocol as-number [route-type] [route-map map-tag]

6.    end

1.    enable

2.    configure terminal

3.    router bgp as-number

4.    no bgp default ipv4-unicast

5.    address-family nsap [unicast]

6.    redistribute protocol [process-id] [route-type] [route-map map-tag]

7.    end

1.    enable

2.    configure terminal

3.    router bgp as-number

4.    no bgp default ipv4-unicast

5.    neighbor peer-group-name peer-group

6.    neighbor peer-group-name remote-as as-number

7.    address-family nsap [unicast]

8.    neighbor peer-group-name route-reflector-client

9.    neighbor ip-address peer-group peer-group

10.    end

1.    enable

2.    configure terminal

3.    router bgp as-number

4.    no bgp default ipv4-unicast

5.    address-family nsap [unicast]

6.    neighbor {ip-address| peer-group-name}prefix-list {clns-filter-expr-name| clns-filter-set-name} in

7.    end

1.    enable

2.    configure terminal

3.    clns filter-set name [deny] template

4.    clns filter-set name [permit] template

5.    router bgp as-number

6.    no bgp default ipv4-unicast

7.    neighbor peer-group-name peer-group

8.    neighbor {ip-address | peer-group-name} remote-as as-number

9.    address-family nsap [unicast]

10.    neighbor {ip-address | peer-group-name} prefix-list {clns-filter-expr-name | clns-filter-set-name} out

11.    neighbor ip-address peer-group peer-group

12.    end

1.    enable

2.    configure terminal

3.    router bgp as-number

4.    no bgp default ipv4-unicast

5.    address-family nsap [unicast]

6.    neighbor {ip-address | peer-group-name} default-originate [route-map map-tag]

7.    end

1.    show clns neighbors

2.    show clns route

3.    show bgp nsap unicast summary

4.    show bgp nsap unicast

Router# show clns neighbors
Tag osi-as-202:
System Id      Interface   SNPA                State  Holdtime  Type Protocol
r1             Se2/0       *HDLC*              Up     274       IS   ES-IS
r3             Et0/1       0002.16de.8481      Up     9         L2   IS-IS
r4             Se2/2       *HDLC*              Up     275       IS   ES-IS

Router# show clns route
Codes: C - connected, S - static, d - DecnetIV
       I - ISO-IGRP,  i - IS-IS,  e - ES-IS
       B - BGP,       b - eBGP-neighbor
C  49.0202.2222 [2/0], Local IS-IS Area
C  49.0202.2222.2222.2222.2222.00 [1/0], Local IS-IS NET
b  49.0101.1111.1111.1111.1111.00 [15/10]
      via r1, Serial2/0
i  49.0202.3333 [110/10]
      via r3, GigabitEthernet0/1/1
b  49.0303.4444.4444.4444.4444.00 [15/10]
      via r4, Serial2/2
B  49.0101 [20/1]
      via r1, Serial2/0
B  49.0303 [20/1]
      via r4, Serial2/2
B  49.0404 [200/1]
      via r9
i  49.0404.9999.9999.9999.9999.00 [110/10]
      via r3, GigabitEthernet0/1/1

Router# show bgp nsap unicast summary
BGP router identifier 10.1.57.11, local AS number 65202
BGP table version is 6, main routing table version 6
5 network entries and 8 paths using 1141 bytes of memory
6 BGP path attribute entries using 360 bytes of memory
4 BGP AS-PATH entries using 96 bytes of memory
0 BGP route-map cache entries using 0 bytes of memory
0 BGP filter-list cache entries using 0 bytes of memory
BGP activity 5/0 prefixes, 8/0 paths, scan interval 60 secs
Neighbor        V    AS MsgRcvd MsgSent   TblVer  InQ OutQ Up/Down  State/PfxRcd
10.1.2.1        4 65101      34      34        6    0    0 00:29:11        1
10.2.3.3        4 65202      35      36        6    0    0 00:29:16        3

Router# show bgp nsap unicast
BGP table version is 3, local router ID is 192.168.3.1
Status codes: s suppressed, d damped, h history, * valid, > best, i - internal,
              r RIB-failure, S Stale
Origin codes: i - IGP, e - EGP, ? - incomplete
   Network          Next Hop            Metric LocPrf Weight Path
*> 49.0101          49.0101.1111.1111.1111.1111.00
                                                           0 65101 i
* i49.0202.2222     49.0202.3333.3333.3333.3333.00
                                                  100      0 ?
*>                  49.0202.2222.2222.2222.2222.00
                                                       32768 ?
* i49.0202.3333     49.0202.3333.3333.3333.3333.00
                                                  100      0 ?
*>                  49.0202.2222.2222.2222.2222.00
                                                       32768 ?
*> 49.0303          49.0303.4444.4444.4444.4444.00
                                                           0 65303 i
*  49.0404          49.0303.4444.4444.4444.4444.00
                                                           0 65303 65404 i
*>i                 49.0404.9999.9999.9999.9999.00
                                                  100      0 65404 i

1.    Attach a console directly to a router running the Cisco IOS XE software release that includes the Multiprotocol BGP (MP-BGP) Support for CLNS feature.

2.    no logging console

3.    Use Telnet to access a router port.

4.    enable

5.    terminal monitor

6.    debug bgp nsap unicast [neighbor-address | dampening | keepalives | updates]

7.    no terminal monitor

8.    no debug bgp nsap unicast [neighbor-address | dampening | keepalives | updates]

9.    logging console

In the following example, the router R1, shown in the figure below, in the autonomous system AS65101 is configured to run BGP and activated to support CLNS. Router R1 is the only Level 2 IS-IS router in autonomous system AS65101, and it has only one connection to another autonomous system via router R2 in AS65202. The no bgp default ipv4-unicast command is configured on the router to disable the default behavior of the BGP routing process exchanging IPv4 addressing information with BGP neighbor routers. After the NSAP address family configuration mode is enabled with the address-family nsap command, the router is configured to advertise the NSAP prefix of 49.0101 to its BGP neighbors and to send NSAP routing information to the BGP neighbor at 10.1.2.2.

router bgp 65101
 no bgp default ipv4-unicast
 address-family nsap
  network 49.0101...
  neighbor 10.1.2.2 activate
  exit-address-family

In the following example, the router R1, shown in he figure below, is configured to run an IS-IS process:

router isis osi-as-101
 net 49.0101.1111.1111.1111.1111.00

The default IS-IS routing process level is used.

In the following example, two of the interfaces of the router R2, shown in the figure below, in the autonomous system AS65202 are configured to run CLNS. GigabitEthernet interface 0/1/1 is connected to the local OSI routing domain and is configured to run IS-IS when the network protocol is CLNS using the clns router isis command. The serial interface 2/0 with the local IP address of 10.1.2.2 is connected with an eBGP neighbor and is configured to run CLNS through the clns enable command:

interface serial 2/0
 ip address 10.1.2.2 255.255.255.0
 clns enable
 no shutdown
!
interface gigabitethernet 0/1/1
 ip address 10.2.3.1 255.255.255.0
 clns router isis osi-as-202
 no shutdown

In the following example, the router R1, shown in the figure below, is configured to advertise the NSAP prefix of 49.0101 to other routers. The NSAP prefix unique to autonomous system AS65101 is advertised to allow the other autonomous systems to discover the existence of autonomous system AS65101 in the network.

router bgp 65101
 no bgp default ipv4-unicast
 neighbor 10.1.2.2 remote-as 64202
 address-family nsap
  network 49.0101...
  neighbor 10.1.2.2 activate

In the following example, the routers R7 and R9, shown in the figure below, in the autonomous system AS65404 are configured to redistribute BGP routes into the IS-IS routing process called osi-as-404. Redistributing the BGP routes allows the Level 2 IS-IS router, R8, to advertise routes to destinations outside the autonomous system AS65404. Without a route map being specified, all BGP routes are redistributed.

router isis osi-as-404
 net 49.0404.7777.7777.7777.7777.00
 redistribute bgp 65404 clns

router isis osi-as-404
 net 49.0404.9999.9999.9999.9999.00
 redistribute bgp 65404 clns

In the following example, the router R2, shown in the figure below, in the autonomous system AS65202 is configured to redistribute Level 2 CLNS NSAP routes into BGP. A route map is used to permit only routes from within the local autonomous system to be redistributed into BGP. Without a route map being specified, every NSAP route from the CLNS level 2 prefix table is redistributed. The no bgp default ipv4-unicast command is configured on the router to disable the default behavior of the BGP routing process exchanging IPv4 addressing information with BGP neighbor routers.

clns filter-set internal-routes permit 49.0202...
!
route-map internal-routes-only permit 10
 match clns address internal-routes
!
router isis osi-as-202
 net 49.0202.2222.2222.2222.2222.00
!
router bgp 65202
 no bgp default ipv4-unicast
 address-family nsap
 redistribute isis osi-as-202 clns route-map internal-routes-only

Router R5, shown in the figure above (in the "Generic BGP CLNS Network Topology" section), has only iBGP neighbors and runs IS-IS on both interfaces. To reduce the number of configuration commands, configure R5 as a member of a BGP peer group called ibgp-peers. The peer group is automatically activated under the address-family nsap command by configuring the peer group as a route reflector client allowing it to exchange NSAP routing information between group members. The BGP peer group is also configured as a BGP route reflector client to reduce the need for every iBGP router to be linked to each other.

In the following example, the router R5 in the autonomous system AS65303 is configured as a member of a BGP peer group and a BGP route reflector client:

router bgp 65303
 no bgp default ipv4-unicast
 neighbor ibgp-peers peer-group
 neighbor ibgp-peers remote-as 65303
 address-family nsap
  neighbor ibgp-peers route-reflector-client
  neighbor 10.4.5.4 peer-group ibgp-peers
  neighbor 10.5.6.6 peer-group ibgp-peers
  exit-address-family

In the following example, the router R1, shown in the figure below, in the autonomous system AS65101 is configured to filter inbound routes specified by the default-prefix-only prefix list:

clns filter-set default-prefix-only deny 49...
clns filter-set default-prefix-only permit default
!
router isis osi-as-101
 net 49.0101.1111.1111.1111.1111.00
!
router bgp 65101
 no bgp default ipv4-unicast
 neighbor 10.1.2.2 remote-as 64202
 address-family nsap
  network 49.0101.1111.1111.1111.1111.00
  neighbor 10.1.2.2 activate
  neighbor 10.1.2.2 prefix-list default-prefix-only in

In the following example, outbound BGP updates are filtered based on NSAP prefixes. This example is configured at Router 7 in the figure below. In this task, a CLNS filter is created with two entries to deny NSAP prefixes starting with 49.0404 and to permit all other NSAP prefixes starting with 49. A BGP peer group is created and the filter is applied to outbound BGP updates for the neighbor that is a member of the peer group.

clns filter-set routes0404 deny 49.0404...
clns filter-set routes0404 permit 49...
!
router bgp 65404
 no bgp default ipv4-unicast
 neighbor ebgp-peers remote-as 65303
 address-family nsap
   neighbor ebgp-peers prefix-list routes0404 out
   neighbor 10.6.7.8 peer-group ebgp-peers

In the figure below, autonomous system AS65101 is connected to only one other autonomous system, AS65202. Router R2 in AS65202 provides the connectivity to the rest of the network for autonomous system AS65101 by sending a default route to R1. Any packets from Level 1 routers within autonomous system AS65101 with destination NSAP addresses outside the local Level 1 network are sent to R1, the nearest Level 2 router. Router R1 forwards the packets to router R2 using the default route.

In the following example, the router R2, shown in the figure below, in the autonomous system AS65202 is configured to generate a default route for router R1 in the autonomous system AS65101, and an outbound filter is created to send only the default route NSAP addressing information in the BGP update messages to router R1.

clns filter-set default-prefix-only deny 49...
clns filter-set default-prefix-only permit default
!
router bgp 65202
 no bgp default ipv4-unicast
 neighbor 10.1.2.1 remote-as 64101
 address-family nsap
  network 49.0202...
  neighbor 10.1.2.1 activate
  neighbor 10.1.2.1 default-originate
  neighbor 10.1.2.1 prefix-list default-prefix-only out

The figure below shows a generic BGP CLNS network containing nine routers that are grouped into four different autonomous systems (in BGP terminology) or routing domains (in OSI terminology). This section contains complete configurations for all routers shown in the figure below.

Figure 3

Components in a Generic BGP CLNS Network

If you need more details about commands used in the following examples, see the configuration tasks earlier in this document and the documents listed in the Additional References.

• Autonomous System AS65101
  
• Autonomous System AS65202
  
• Autonomous System AS65303
  
• Autonomous System AS65404