© 2005 Cisco Systems, Inc. All rights reserved. BGP v3.2—5-1 Customer-to-Provider Connectivity with BGP Connecting a Multihomed Customer to a Single Service Provider

© 2005 Cisco Systems, Inc. All rights reserved. BGP v3.2—5-2 Outline Overview Configuring BGP on Multihomed Customer Routers Conditional Advertising in Multihomed Customer Networks Configuring BGP on Service Provider Routers Removing Private AS Numbers BGP Support for Dual AS Configuration for Network AS Migrations Backup Solutions with BGP Load Sharing with the Multihomed Customer Load Sharing with BGP Multipath Load Sharing with EBGP Multihop Summary

© 2005 Cisco Systems, Inc. All rights reserved. BGP v3.2—5-3 Configuring BGP on Multihomed Customer Routers

© 2005 Cisco Systems, Inc. All rights reserved. BGP v3.2—5-4 The customer address space is advertised on every customer edge router. Customer edge routers run IBGP between themselves and advertise the default route to the rest of the customer network. Configuring BGP on Multihomed Customer Routers (Cont.)

© 2005 Cisco Systems, Inc. All rights reserved. BGP v3.2—5-5 Customer edge routers should announce the whole customer address space into BGP. The static route covering the whole customer address should point to the core of the customer network, not to null 0. The customer edge router revokes the BGP announcement of customer address space if the edge router loses connectivity with the customer core. Conditional Advertising in Multihomed Customer Networks

© 2005 Cisco Systems, Inc. All rights reserved. BGP v3.2—5-6 Conditional Advertising in Multihomed Customer Networks (Cont.)

© 2005 Cisco Systems, Inc. All rights reserved. BGP v3.2—5-7 Configuring BGP on Service Provider Routers The service provider must: Advertise a default route to the customer through BGP Filter incoming BGP updates with a prefix-list to verify that the customer announces only the assigned address space Filter incoming BGP updates with an AS-path filter-list to verify that the customer uses only its own AS number Optionally, the no-export community should be set on customer routes.

© 2005 Cisco Systems, Inc. All rights reserved. BGP v3.2—5-8 Configuring BGP on Service Provider Routers (Cont.) neighbor ip-address default-originate router(config-router)# By default, the default route ( /0) is not advertised in outgoing BGP updates. The neighbor default-originate command advertises the default route to a BGP neighbor even if the default route is not present in the BGP table. Note: The default route is not sent through the outbound BGP filters (prefix-list, filter-list, or route-map).

© 2005 Cisco Systems, Inc. All rights reserved. BGP v3.2—5-9 Configuring BGP on Service Provider Routers (Cont.)

© 2005 Cisco Systems, Inc. All rights reserved. BGP v3.2—5-10 Private AS Number Propagation Private AS numbers should not be advertised into the Internet. The private AS numbers must be removed from the AS path before the customer BGP routes are advertised to other service providers.

© 2005 Cisco Systems, Inc. All rights reserved. BGP v3.2—5-11 Removing Private AS Numbers neighbor ip-address remove-private-as router(config-router)# The command modifies AS-path processing on outgoing updates sent to specified neighbor. Private AS numbers are removed from the tail of the AS path before the update is sent. Private AS numbers followed by a public AS number are not removed. The AS number of the sender is prepended to the AS path after this operation.

© 2005 Cisco Systems, Inc. All rights reserved. BGP v3.2—5-12 Removing Private AS Numbers (Cont.)

© 2005 Cisco Systems, Inc. All rights reserved. BGP v3.2—5-13 BGP Support for Dual AS Configuration for Network AS Allows you to merge a secondary AS under a primary AS without disrupting customer peering sessions Allows a router to appear, to external peers, as a member of secondary AS during the AS migration Allows a network operator to merge the autonomous systems and then later migrate customers to new configurations during normal service windows without disrupting existing peering arrangements If misconfigured, increases the possibility that routing loops can be created

© 2005 Cisco Systems, Inc. All rights reserved. BGP v3.2—5-14 Backup Solutions with BGP The route selection is controlled entirely by the customer routers. Local preference is used to differentiate primary and backup links for the outgoing traffic. The MED is used to differentiate primary and backup links for the return traffic. No service provider configuration is required.

© 2005 Cisco Systems, Inc. All rights reserved. BGP v3.2—5-15 Primary and Backup Link Selection

© 2005 Cisco Systems, Inc. All rights reserved. BGP v3.2—5-16 Load sharing of outgoing customer traffic is identical to the static routing scenario. You can implement load sharing of return traffic in a number of ways: Announce portions of the customer address space to each upstream router Configure BGP multipath support in the service provider network Use EBGP multihop in environments where parallel links run between a pair of routers Load Sharing with the Multihomed Customer

© 2005 Cisco Systems, Inc. All rights reserved. BGP v3.2—5-17 maximum-paths number router(config-router)# By default, BGP selects a single path as the best path and installs it in the IP routing table. With maximum-paths configured, a BGP router can select several identical EBGP routes as the best routes and install them in the IP routing table for load-sharing purposes. The BGP router can install up to six BGP routes in the IP routing table. Configuring BGP Multipath Support

© 2005 Cisco Systems, Inc. All rights reserved. BGP v3.2—5-18 Because of recursive lookup, load sharing toward a BGP destination always occurs if there are several equal-cost IGP paths to the BGP next hop. Equal-cost IGP paths are easily generated if the BGP next hop is not directly connected. Load Sharing with EBGP Multihop

© 2005 Cisco Systems, Inc. All rights reserved. BGP v3.2—5-19 Configuring Multihop EBGP Sessions neighbor ip-address ebgp-multihop [ TTL ] router(config-router)# By default, EBGP neighbors must be directly connected. The ebgp-multihop command declares an EBGP neighbor to be distant (several hops away). The number of hops can be specified in the TTL parameter. This command is usually used to run EBGP between loopback interfaces for dial backup or load-sharing purposes. Use with extreme caution; routing loops can occur very easily.

© 2005 Cisco Systems, Inc. All rights reserved. BGP v3.2—5-20 Load Sharing with EBGP Multihop (Cont.)

© 2005 Cisco Systems, Inc. All rights reserved. BGP v3.2—5-21 Summary When a customer has multiple connections to a single ISP and the link-level procedures cannot detect a link failure, a routing protocol is required. For security reasons, this routing protocol must be BGP. The AS number that is used by the customer does not have to be a public AS number; it can be a private AS number in the range to When conditionally advertising customer networks to the ISP, you should use a static route covering the whole customer address space and pointing to the core of the customer network instead of null 0. The ISP should advertise a default route to the customer through BGP. Incoming filters should also be used by the provider to ensure that only the correct address space and AS number are advertised by the customer. The BGP Support for Dual AS Configuration for Network AS Migrations feature allows you to merge a secondary AS under a primary AS without disrupting customer peering sessions.

© 2005 Cisco Systems, Inc. All rights reserved. BGP v3.2—5-22 Summary (Cont.) Private AS numbers must never be propagated to the rest of the Internet. The ISP must therefore remove the private AS numbers from the AS path before sending them to another public AS. You can use parallel links between the customer network and the network of a single ISP for backup or load-sharing purposes. The customer can control the outgoing load using local preference and also control the incoming load using the MED (metric) attribute. With the MED, the links go to a single remote AS. By announcing portions of its address space, a customer can use maximum paths and EBGP multihop to provide load sharing over multiple links. EBGP multihop can be used for load balancing only if redundant links terminate on the same provider router.

© 2005 Cisco Systems, Inc. All rights reserved. BGP v3.2—5-23