Treating Missing Meds As The Worst Meds; Configuring Route Reflection Parameters - Foundry Networks FESX Manual

Foundry Configuration Guide for the FESX, FSX, and FWSX

Treating Missing MEDs as the Worst MEDs

By default, the Layer 3 Switch favors a lower MED over a higher MED during MED comparison. Since the Layer 3
Switch assigns the value 0 to a route path's MED if the MED value is missing, the default MED comparison results
in the Layer 3 Switch favoring the route paths that are missing their MEDs.
To change this behavior so that the Layer 3 Switch favors a route that has a MED over a route that is missing its
MED, enter the following command at the BGP4 configuration level of the CLI:
FESX424 Router(config-bgp-router)# med-missing-as-worst
Syntax: [no] med-missing-as-worst
NOTE: This command affects route selection only when route paths are selected based on MED comparison. It
is still possible for a route path that is missing its MED to be selected based on other criteria. For example, a route
path with no MED can be selected if its weight is larger than the weights of the other route paths.

Configuring Route Reflection Parameters

Normally, all the BGP routers within an AS are fully meshed. Each of the routers has an IBGP session with each
of the other BGP routers in the AS. Each IBGP router thus has a route for each of its IBGP neighbors. For large
ASs containing many IBGP routers, the IBGP route information in each of the fully-meshed IBGP routers can
introduce too much administrative overhead.
To avoid this problem, you can hierarchically organize your IGP routers into clusters.
• A cluster is a group of IGP routers organized into route reflectors and route reflector clients. You configure
the cluster by assigning a cluster ID on the route reflector and identifying the IGP neighbors that are members
of that cluster. All the configuration for route reflection takes place on the route reflectors. The clients are
unaware that they are members of a route reflection cluster. All members of the cluster must be in the same
AS. The cluster ID can be any number from 1 – 4294967295. The default is the router ID, expressed as a 32-
bit number.
NOTE: If the cluster contains more than one route reflector, you need to configure the same cluster ID on all
the route reflectors in the cluster. The cluster ID helps route reflectors avoid loops within the cluster.
• A route reflector is an IGP router configured to send BGP route information to all the clients (other BGP4
routers) within the cluster. Route reflection is enabled on all Foundry BGP4 routers by default but does not
take effect unless you add route reflector clients to the router.
• A route reflector client is an IGP router identified as a member of a cluster. You identify a router as a route
reflector client on the router that is the route reflector, not on the client. The client itself requires no additional
configuration. In fact, the client does not know that it is a route reflector client. The client just knows that it
receives updates from its neighbors and does not know whether one or more of those neighbors are route
reflectors.
NOTE: Route reflection applies only among IBGP routers within the same AS. You cannot configure a cluster
that spans multiple ASs.
Figure 21.2 shows an example of a route reflector configuration. In this example, two Layer 3 Switches are
configured as route reflectors for the same cluster. The route reflectors provide redundancy in case one of the
reflectors becomes unavailable. Without redundancy, if a route reflector becomes unavailable, its clients are cut
off from BGP4 updates.
AS1 contains a cluster with two route reflectors and two clients. The route reflectors are fully meshed with other
BGP4 routers, but the clients are not fully meshed. They rely on the route reflectors to propagate BGP4 route
updates.
21 - 32 © Foundry Networks, Inc. December 2005