3Com Router User Manual

							Configuring BGP 457
To configure an advanced BGP peer group configuration:
1Configure the AS number of BGP peer group
Ta b l e 528   Configure AS Number of BGP Peer Group
By default, there is no AS number for BGP peer group.
2Configure connection between peers indirectly connected
Ta b l e 529   Configure Connection Between Peers Indirectly Connected
By default, it only allows direct-connection peer.
The maximum hop value is ttl. The default value is 64, ranging from 1 to 255.
3Set the timers of BGP peer group
Ta b l e 530   Set the Timers of BGP Peer Group
By default, the interval of sending keepalive packet is 60 seconds, the interval of 
holdtime is 180 seconds,
Note that the timers configured with this command are of higher preference than 
the values configured with the 
timers command.
4Configure the BGP routing update sending interval
Ta b l e 531   Configure BGP Routing Update Sending Interval
By default, the BGP routing update sending interval is 5 seconds
5Configure to send the community attribute to a BGP peer group
Ta b l e 532   Configure to Send Community Attribute to a BGP Peer Group
OperationCommand
Configure AS number of BGP peer grouppeer group-name as-number as-number
Remove AS number of BGP peer groupundo peer group-name as-number as-number
OperationCommand
Configure connection between peers 
indirectly connectedpeer group-name ebgp-max-hop [ ttl ]
Return to the default BGP connections to 
external peer groupundo peer group-name ebgp-max-hop
OperationCommand
Set the timers of BGP peer grouppeer group-name timers keepalive-interval 
holdtime-interval
Restore the timers of BGP peer group to 
default valueundo peer group-name timers
OperationCommand
Configure BGP routing update sending 
intervalpeer group-name 
route-update-interval seconds
Restore BGP routing update sending 
intervalundo peer group-name 
route-update-interval
OperationCommand
Configure to send community attribute to 
a BGP peer grouppeer group-name advertise-community
Delete the BGP community en to the peer 
group.undo peer group-name 
advertise-community 
						

							458CHAPTER 30: CONFIGURING BGP
By default, send no community attribute to any peer group.
6Configure a peer group as the client of a BGP reflector
In general, the AS requires that all the IBGP routers should be connected to one 
another, and the routes sent by the IBGP neighbors is not advertised, to prevent 
route loop. However, if the route reflector is used, not all IBGP speakers are 
required to be fully connected. This technique requires configuring an internal BGP 
peer as a router reflector. Other internal peers are not necessarily mesh connected 
but set up an IBGP session with the route reflector and learn routes through the 
route reflector. Using 
peer reflect-client command, you can configure 
internal neighbors which can communicate with the route reflector. These 
neighbors are the client-group members of the route reflector. Other neighbors 
are the non-client-group members.
Generally, it is unnecessary to configure this command for the peer entity since the 
IBGP peer is in its default group. You should use the 
peer peer-address 
reflect-client
 command to configure the route reflector client.
Ta b l e 533   Configure Peer Group as the Client of BGP Reflector
7Configure to send the default route to the peer group
Ta b l e 534   Configure to Send the Default Route to the Peer Group
By default, the local router does not advertise the default route to any peer group. 
A next hop should be sent to the peer unconditionally as the default route.
8Set the router’s own IP address as the next hop when the peer group distributes 
route information.
Cancel the processing of next hop when sending a route to the peer and take the 
self-address as the next hop.
Ta b l e 535   Set the Own IP address as Next Hop When Peer Group Distributes Route
By default, the router’s own IP address is not set as the next hop when the peer 
group distributes routes.
9Create a routing policy for the peer group 
Ta b l e 536   Create Routing Policy for Peer Group
OperationCommand
Configure peer group as the client of BGP 
reflectorpeer group-name reflect-client
Disable peer group as the client of BGP 
reflectorundo peer group-name reflect-client
OperationCommand
Configure to send the default route to 
peer grouppeer group-name 
default-route-advertise
Do not allow to send default route to the 
peersundo peer group-name 
default-route-advertise
OperationCommand
Set the own IP address as next hop when 
peer group distributes routepeer group-name next-hop-local
Not to set the own IP address as next hop 
when peer group distributes routeundo peer group-name next-hop-local
OperationCommand 
						

							Configuring BGP 459
By default, the route from the peer or peer group is not designated with any route 
policy.
10Create a filtering policy based on the access list for the peer group
Ta b l e 537   Create a Filtering Policy Based on Access List for Peer Group
By default, no route filtering policy based on IP ACL for peer group is set.
11Create a BGP route filtering based on the AS path for the peer group
Ta b l e 538   Create a BGP Route Filtering Based on AS Path for Peer Group
By default, a BGP filtering is disabled.
12Configure the BGP version of peer group 
Ta b l e 539   Configure BGP Version of Peer Group
By default, software accepts BGP Version 4.
Creating Aggregate 
AddressesCIDR supports manual route aggregation. Manual aggregation, using the 
aggregate command adds a piece of routing aggregate information to the BGP 
routing table. The parameters can be set at the same time when manual 
aggregation mode is configured.
Perform the following configurations in BGP view.
Ta b l e 540   Create an Aggregate Addresses
Create routing policy for peer grouppeer group-name route-policy 
policy-name { import | export }
Remove a routing policy to import or 
export routesundo peer group-name route-policy 
policy-name { import | export }
OperationCommand
Create an filtering policy based on access 
list for peer grouppeer group-name filter-policy 
acl-number { import | export }
Delete an filtering policy based on access 
list for peer groupundo peer group-name filter-policy 
acl-number { import | export }
OperationCommand
Create a BGP route filtering based on AS 
path for peer grouppeer group-name acl 
aspath-list-number { import | export 
}
Delete a BGP route filtering based on AS 
path for peer groupundo peer group-name acl 
aspath-list-number { import | export 
}
OperationCommand
Configure the BGP version of peer grouppeer group-name version 
version-number
Restore the default BGP version for the 
peer groupundo peer group-name version
OperationCommand 
						

							460CHAPTER 30: CONFIGURING BGP
By default, an aggregate is disabled.
Configure BGP Route 
ReflectorTo guarantee the connectivity between the IBGP peers, an all-closed network 
should exist between IBGP peers. In some networks, the internal BGP network can 
become very large (with more than one hundred sessions in each router), resulting 
in huge overhead. The route reflector designates a central router as the core of the 
internal sessions. Multiple BGP routers can become peers with this central router, 
and then multiple route reflectors can be peers with each other.
Since the route reflector is the center of other routers, other routers are called 
client routers from the viewpoint of the reflector. The client routers are peers of 
the route reflector and exchange routing information. The route reflector forwards 
(reflects) information among the client routers in turn.
As shown in the following diagram, Router A receives an update from an external 
peer and transfers it to Router B. Router B is a route reflector, which has two 
clients: Router A and Router C.
Router B can reflect the routing update from client Router A to client Router C. In 
this instance, the session between Router A and Router C is unnecessary because 
the route reflector forwards the BGP information to Router C.
Figure 152   Schematic diagram of route reflector
The route reflector divides the IBGP peers into two types: client and non-client. 
Using the peer reflect-client command, you can configure the internal neighbors 
that can communicate with the route reflector. The neighbors are called the client 
group members of the route reflector, and other neighbors that are not configured 
as the non-clients are the non-client group members of the route reflector.
Add a piece of routing aggregate 
information to the BGP routing tableaggregate address mask [ as-set ] [ 
detail-suppressed  ] [ 
suppress-policy policy-name ] [ 
origin-policy policy-name ] [ 
attribute-policy policy-name ]
Delete a piece of routing aggregate 
information to the BGP routing tableundo aggregate address mask
EBGP EBGPRoute Reflector
Reflected router
Update routeRouter B
Router A
Router C 
						

							Configuring BGP 461
The non-clients must form an all-closed network with the reflector, as they follow 
the basic rules of IBGP. A client should not be peer of other internal speakers 
outside its cluster. The reflecting function is achieved only on the route reflector. 
All the clients and non-clients are normal BGP peers irrelevant to the function. A 
client is a client only because the route reflector regards it as the client.
When the router reflector receives several routes to one destination, it chooses the 
best one based on the usual BGP routing strategy process. The best route transfers 
inside AS according to following rules:
■If the route is received from non-client peers, it only reflects to clients.
■If the route is received from client peers, it reflects to all the clients and 
non-clients except this routes sender.
■If an EBGP peer receives the route, it is reflected to all clients and non-client 
peers that can be reflected.
1Configure the route reflection between clients.
Perform the following configurations in BGP view.
Ta b l e 541   Configure the Route Reflection Between Clients
By default, the route reflection function is disabled between the clients within the 
reflection group. 
Note that the route reflector configuration between the clients is invalid if the 
clients are fully connected.
2Configure the cluster ID.
As the route reflector is imported, the route selection circle can occur in an AS, 
and the route that leaves a cluster during update may try to reenter this cluster. 
The traditional AS routing method cannot detect the internal circle of the AS, 
because the update has not left the AS yet. BGP provides two methods to avoid an 
AS internal loop when you configure the route reflector:
aConfigure an originator-ID for the route reflector:
The originator-ID is a 4-bit, optional, non-transitional BGP attribute created by 
the route reflector. It carries the router ID of the originator. If the configuration 
is improper, and the routing update returns to the originator, the originator will 
discard it.
You dont need to configure this parameter, and it functions automatically 
when the BGP protocol is started.
bConfigure the cluster-ID of the route reflector:
Generally, a cluster has only one route reflector. To avoid routing update 
information failure due to the route reflector failure, multiple route reflectors 
are recommended for a cluster. If more than one route reflector exists in a 
cluster, all the route reflectors must be configured with the same cluster ID.
Perform the following configurations in BGP view.
OperationCommand
Enable route reflection function between 
the clients within the reflection groupreflect between-clients
Disable route reflection function between 
the clients within the reflection groupundo reflect between-clients 
						

							462CHAPTER 30: CONFIGURING BGP
Ta b l e 542   Configure the Cluster ID
By default, the router ID of the route reflector is used as the cluster ID.
Configuring a BGP 
CommunityIn BGP range, a community is a logical area formed by a group of destinations 
which share common attributes for applying the route policy. A community is not 
limited to a network or an AS, and has no physical boundary.
The community attribute is an optional and transitional attribute. Some 
communities are commonly recognized and globally functional. These 
communities are called standard communities. Sometimes the extended 
community attribute can be defined for special purposes.
The community attribute list is used to identify the community information. It can 
be a standard-community-list and an extended-community-list.
In addition, one route can have more than one community attribute. The speaker 
with multiple community attributes in a route can work according to one, several 
or all attributes. The community attribute can be added or modified before the 
router transfers a route to other peers.
Perform the following configurations in system view.
Ta b l e 543   Configure BGP Community
By default, no community list is created.
Configuring a BGP AS 
Confederation AttributeConfederation is another method to solve the problem of a sudden increase of 
IBGP closed networks inside an AS. An AS is divided into multiple sub-ASs and the 
IBGP peers inside the sub-ASs are fully connected, and each sub-AS connects with 
other sub-ASs inside the confederation. Among the subsystem, the peers perform 
EBGP sessions, but they can exchange routing information just like IBGP peers. All 
the important information such as the next hop, MED value and the local priority 
will not be lost when passing through the AS. 
OperationCommand
Configure Cluster-ID of the 
Route-Reflectorreflect cluster-id cluster-id 
Remove Cluster-ID of the Route-Reflectorundo reflect cluster-id cluster-id 
OperationCommand
Create a standard-community-listip community-list 
standard-community-list-number { 
permit | deny } { aa:nn | internet | 
no-export-subconfed | no-advertise | 
no-export }
Specify a extended-community-list entryip community-list 
extended-community-list-number { 
permit | deny } as-regular-expression
Delete the specified community listundo ip community-list 
{standard-community-list-number | 
extended-community-list-number } 
						

							Configuring BGP 463
The disadvantage is that when a non-confederation scheme changes to a 
confederation scheme, it is required to reconfigure the router and to modify the 
logical topology. In addition, if the BGP strategy is not manually configured, the 
best path may not be selected through the confederation.
1Configure a Confederation
You can use different IGP for each sub-AS. Externally, a sub-AS is an integer and 
the confederation ID is the identification of the sub-AS.
Perform the following configurations in BGP view.
Ta b l e 544   Configure a Confederation
By default, no BGP confederation identifier is specified.
2Configure the sub-system of e confederation
The configured sub-AS is inside a confederation and each sub-AS uses fully closed 
network. Use 
confederation id command to specify the confederation ID of the 
AS. If the confederation ID is not configured, this configuration item is invalid.
Perform the following configurations in BGP view.
Ta b l e 545   Configure the Sub-system of E Confederation
By default, no confederation peers are specified.
3Configure the non-RFC standard AS confederation attributes.
The creation of an AS confederation in the devices from some other providers may 
not be consistent with the RFC1965 standard. All the routers in the confederation 
must be configured as using non-RFC1965 standard AS confederation attributes 
to create interconnections with the router using non-RFC1965 standard AS 
confederation. 
Perform the following configurations in BGP view.
Ta b l e 546   Configure to Use the Non-RFC Standard AS Confederation Attributes.
By default, 3Com routers use the RFC1965 standard AS confederation attributes.
OperationCommand
Specify a Confederation idconfederation id as-number
Remove a Confederation idundo confederation id
OperationCommand
Configure the sub-system of e 
confederationconfederation peer-as as-number [ 
as-number ] …
Delete an AS from the confederationundo confederation peer-as as-number 
[ as-number ] …
OperationCommand
Configure the non-RFC standard AS 
confederation attributes.confederation nonstandard-compatible
Remove the non-RFC standard AS 
confederation attributes.undo confederation 
nonstandard-compatible 
						

							464CHAPTER 30: CONFIGURING BGP
Configuring Route 
DampeningRoute instability is frequently indicated when a route disappears that used to exist 
in the routing table. This route may reappear and disappear frequently, which is 
called routing flapping. When there is route flapping, the UPDATE and 
WITHDRAWN messages are broadcast repeatedly over the network, occupying 
bandwidth and processing time of the routers. The administrator should take 
action to prevent route flapping. Route dampening is a technology to control 
routing flapping.
There are two types of routes, stable routes and unstable routes. Stable routes 
remain in the route table continuously, while unstable routes should be suppressed 
by route dampening. The unstable route is penalized by not allowing it to 
advertise when its penalty level reaches a threshold. The penalty is exponentially 
decreased as time goes by. Once it is lower than a certain threshold, the route is 
unsuppressed and is advertised again, as shown in the following diagram.
Figure 153   Schematic diagram of route dampening
Configure the following parameters to adjust the performance of route 
dampening:
■Penalty: Increases upon each route flap, decays as time goes by.
■Reachable-half-life: Time duration before they reachable route penalty is 
reduced to half.
■Unreachable-half-time: Time duration before the unreachable route penalty 
is reduced to half.
■Ceiling-max-suppress: The maximum value of the penalty.
■Suppress-limit: The route advertisement is suppressed when the penalty 
reaches this threshold.
■Reuse-limit: The route advertisement is unsuppressed when the penalty is 
lower than this value.
1Configure route dampening.
Perform the following configurations in BGP view.
The parameters are mutually dependent. To configure any parameter, all other 
parameters should also be specified.
Suppression threshold
Threshold to reuse
Time Penalty 
						

							Configuring BGP 465
Ta b l e 547   Configure Route Dampening
By default, route dampening is disabled.
2Display route flap information.
Perform the following configurations in system view.
Ta b l e 548   Display Route Flap Information
Configuring 
Synchronization of BGP 
and IGPBGP protocol prescribes that a BGP router does not advertise the destination 
known through internal BGP peers to external peers unless the destination can be 
known also through IGP. If a router can know the destination through IGP, then 
the route can be distributed in the AS because an internal connection has been 
ensured.
One major task of the BGP protocol is to distribute the network reachable 
information of the local AS to other ASs. Therefore, BGP needs to distribute the 
route information by synchronization with IGP (such as RIP and OSPF), 
Synchronization means that BGP cannot distribute transition information to other 
ASs until IGP broadcasts the route information successfully within its AS. That is to 
say, before a router receives an updated destination information from an IBGP 
OperationCommand
Put BGP route attenuation in effect or 
modify BGP route attenuation parameterdampening [ half-life-reachable 
half-life-unreachable reuse suppress 
ceiling ] [ route-policy policy-name 
]
Clear route routing dampening 
information and de-suppress the 
suppressed routereset dampening [ network-address [ 
mask ] ]
Disable the route dampeningundo dampening
OperationCommand
Display BGP dampened routesdisplay bgp routing-table dampened
Display flap information of all routes display bgp routing-table 
flap-information
Display the route flap statistics of routes 
with AS path comply with regular 
expressiondisplay bgp routing-table 
flap-information regular-expression 
as-regular-expression
Reset BGP flap information matching AS 
path regular expressionreset bgp flap-information 
regular-expression [ 
as-regular-expression ]
Display the route flap statistics of routes 
that passed AS filter-listdisplay bgp routing-table 
flap-information acl 
[aspath-list-number ]
Clear BGP flap information matching the 
specified filter listreset bgp flap-information acl 
[aspath-list-number ]
Display the route flap statistics of routes 
with designated destination addressdisplay bgp routing-table 
flap-information network-address 
mask [ longer-match ]
Clear the route flap statistics of routes 
with designated destination addressreset bgp flap-information 
network-address
Clear the route flap statistics of routes 
received from the specified peer.reset bgp network-address 
flap-information 
						

							466CHAPTER 30: CONFIGURING BGP
peer and advertises it to other EBGP peers, it will try to check whether this 
destination can be reached through its AS.
Perform the following configurations in BGP view.
Ta b l e 549   Configure Synchronization of BGP and IGP
By default, BGP synchronizes with IGP.
3Com routers provide the ability to cancel BGP and IGP synchronization so the 
route from IBGP can be distributed without continuously checking if the IGP route 
still exists.
The synchronization of a border router can be shut down safely in the following 
cases:
■All the routers of an AS can form an IBGP totally-closed network. In such a 
case, a route known from any border routers EBGP can be automatically 
transferred to any other router through IBGP so that the connection of the AS 
is insured.
■When AS is not a transitional AS.
Configuring the 
Interactions between 
BGP and an IGP
 
BGP can import route information that is found by running IGP in another AS to its 
own AS.
Perform the following configurations in BGP view.
Ta b l e 550   Configure Route Import for BGP
By default, BGP does not import routes from other domains into the routing table.
The protocol attribute specifies the source routing domain that can be imported. 
At present, BGP can import routes domain such as connected, static, RIP, OSPF and 
OSPF-ASE.
See “Configure Route Import” in “Configuration of IP Routing Policy” for the 
details of routing import.
The import-route command cannot import the default route into BGP, so you 
must use the 
default-information command to import the default route into 
BGP.
Perform the following configurations in BGP view.
OperationCommand
Synchronize BGP with IGPsynchronization
Prohibit synchronizing BGP with IGPundo synchronization
OperationCommand
Configure route import for BGPimport-route protocol [ med med ] [ 
route-policy policy-name ]
Cancel route distribution for BGPundo import-route protocol