Advanced BGP – Chapter Summary

The following section is a summary of the major points you should be aware of in this chapter.

Border Gateway Protocol Path Control Overview

• It is important to remember that BGP is first and foremost a routing policy control tool
• BGP attributes can be used to influence both inbound and outbound routing
• In Cisco IOS software, the BGP path control policies are implemented using route maps
• When BGP policies are implemented or changed they do not take effect immediately
• Use the clear ip bgp command to immediately apply the configuration changes

Influencing Inbound Path Selection

• There are BGP two attributes that are used to influence the inbound path:

1. The MULTI_EXIT_DISC attribute
2. The AS_PATH attribute

• The MED should be used when multihomed to the same neighboring AS
• The MED is not compared for routes from different ASes
• The received MED is not re-advertised to external peers
• The AS_PATH should be used when multihomed to more than a single AS

Influencing Outbound Path Selection

• There are two attributes  that are used to influence the outbound path:

1. The LOCAL_PREF Attribute
2. The WEIGHT Attribute

• The LOCAL_PREF is ignored in UPDATE messages received from external peers
• The LOCAL_PREF is included in UPDATE messages sent to internal peers
• The WEIGHT is a Cisco-proprietary attribute
• The WEIGHT is a locally significant attribute that affects only the local BGP speaker

Basic BGP Load Balancing and Load Sharing

• Load balancing is used to split the load to the same destination across multiple paths
• Load balancing distributes traffic evenly across multiple links
• Load balancing is unidirectional
• Load balancing is supported at Layers 2, 3 and 4
• Load sharing is also used to split the load to the same destination across multiple paths
• Load sharing does not distribute the traffic evenly across the multiple paths
• Load sharing can be used to distribute both incoming and outgoing network traffic

Border Gateway Protocol Peer Groups

• Peers groups are used to simplify the BGP configuration for internal and external peers
• Members in the peer group must have common characteristics
• Peer groups are also used to optimize BGP by reducing the convergence time
• All peer groups have a peer group leader
• The members in the peer group must be synched to the leader
• The local BGP speaker generates an UPDATE for the leader and replicates it to members

Border Gateway Protocol Route Filtering

• Cisco IOS software supports many options that can be used for BGP routing policy control
• BGP routing policy control tools include the following:

1. IP Prefix Lists
2. BGP Communities
3. AS_PATH Filters

• For BGP, prefix lists can be applied in two ways:

1. Using the neighbor [address] prefix-list [name] <in|out> command
2. Using the neighbor [address] route-map [name] <in|out> command

• For BGP, the well-known communities can be assigned to prefixes using route maps
• To include communities, the neighbor send-community command must be used
• AS path filters are based on regular expressions and match on the AS_PATH
• A regular expression is a pattern to match against an input string
• A regular expression comprises four main parts. These four main parts are:

1. A Range
2. An Atom
3. A Piece
4. A Branch

Border Gateway Protocol Aggregation

• There are two ways BGP can be configured to perform route summarization:

1. Using the network [network] mask [mask] command
2. Using the aggregate-address [network][mask] command

• The aggregate-address command allows specific entries to be advertised by default
• The summary-only keyword suppresses specific entries when aggregating

BGP and IGP Route Redistribution

• There is no special configuration when redistributing EIGRP routes into BGP
• Special considerations for redistributing OSPF into BGP include:

1. By default, all of the internal OSPF routes, i.e. routes marked O and IA, are redistributed
2. By default, external Type 1 OSPF routes, i.e. routes marked E1, are redistributed
3. By default, external Type 2 OSPF routes, i.e. routes marked E2, are NOT redistributed

• To redistribute external Type 2 routes, you must use the match external command
• BGP routes should NEVER be redistributed into an IGP without filtering
• By default, only external BGP routes are imported into the IGP during redistribution
• The bgp redistribute-internal command is used to redistribute iBGP routes

BGP Route Reflectors and Confederations

• By default, an iBGP peer will not advertised iBGP-received routes to another iBGP peer
• The default iBGP behavior means a full mesh is required to allow route connectivity
• There are two features that can be used to negate the full mesh internal BGP requirement:

1. Route Reflectors
2. Confederations

• Route reflection works via the use of a Route Reflector or Route Reflector routers
• With RRs, iBGP routers are classified into three different router groups:

1. Route Reflectors (RR)
2. Regular iBGP speakers (non-clients)
3. Route Reflector Clients

• Route Reflectors are BGP speakers that reflect routes between the following BGP speakers:

1. Between clients and non-clients
2. Between clients and clients (called client-to-client reflection)
3. Between non-clients and clients

• UPDATES received from the RR include the ORIGINATOR_ID and CLUSTER_LIST
• The ORIGINATOR_ID contains the RID of the router that generated the UPDATE
• The CLUSTER_LIST contains the CLUSTER_ID set by the local RR
• These two additional attributes are not included in UPDATES sent to external peers
• Confederations are used to split the AS into sub-ASes
• Peers in different sub-ASes establish external iBGP (eiBGP) peer sessions
• UPDATES for prefixes from a router in an external sub-AS are CONFED-EXTERNAL
• UPDATES for prefixes from a router in the same sub-AS are CONFED-INTERNAL
• The sub-ASes traversed are printed in parenthesis (brackets)
• The sub-ASes are not included in UPDATES to external, non-confederation peers

Border Gateway Protocol Dampening

• BGP route flap dampening, as defined in RFC 2439 has three major goals, which are:

1. It provides a mechanism to reduce router processing load caused by unstable routes
2. It prevents sustained route fluctuations, i.e. routes going up and down
3. It provides stability without sacrificing convergence time for well-behaved routes

• Route dampening is used to penalize route or prefix entries that are repeatedly flapping
• The bgp dampening router command is used to enable route flap dampening

Border Gateway Protocol Authentication

• MD5 authentication is used to secure or verify the security of the TCP segments
• MD5 authentication requires the use of a shared password between peers
• The shared password is configured using the neighbor…password command
• When configuring the shared password, the following should be taken into consideration:

1. The password is a case-sensitive string that may be up to 25 characters in length
2. The string can contain any alphanumeric characters, including spaces
3. A password cannot be configured with a leading number, followed by a space