Wednesday, November 29, 2006

Implementing BGP in Large Networks

1.If a route reflector hears an update from a nonclient, what action will be taken?

** If a route reflector hears an update from a nonclient, it will reflect the update to clients only.

2: In version 11.0 of the Cisco IOS software, what method would be used to restrict routing information from being received or propagated?

** Distribute lists would be used to restrict BGP routing updates in version 11.0 of the Cisco IOS software. Prefix lists became available to ISPs in 11.2 of the IOS and in 12.0 to the general public.

3:Explain the purpose and use of the command show ip prefix-list name [seq seq-number].

** This command displays the entry prefix list with a given sequence number.

4: Why would you redistribute static routes into BGP?

** The static routes are forwarded to the nonexistent interface null 0 to aggregate the routes to create a supernet. This is then redistributed to the BGP process instead of the hundreds of routes that exist. Because any route redistributed into BGP must be in the IP routing table, this is a way of creating a supernet and having it redistributed.
A static route to null 0 is not needed if the network command is used and no address aggregation is performed, although it can be used to prevent routing loops.

5: Why is it advisable to have the route reflectors fully meshed?

** Although clients are not fully meshed within a cluster, it is important that the route reflectors they serve are fully meshed. This is to ensure that the routers pass routing information to each other. Remember that iBGP does not propagate routing information; it generates information only based on the network command. The route reflectors need to pass information to each other to ensure the entire network has a full knowledge of the network. It is possible to create a hierarchical design with route reflectors, but this should be done with great caution, as it can cause routing loops.

6: Why is filtering often required when redistributing BGP into an IGP?

** Filtering is often required when redistributing BGP into an IGP because the routing tables can become overwhelmed by the number of routes that are imported.

7: What are the advantages of multihoming?

** Multihoming allows for redundancy, which is important when the link is into the Internet and can carry crucial business information for the company. Multihoming also increases performance by allowing the selection of the better paths.

8: Why do iBGP peers need to be fully meshed?

** iBGP needs to be fully meshed because it has a fundamental rule stating that iBGP will not propagate back into iBGP a route that it has learned from iBGP. The reason is that this is the only way to prevent routing loops. If each router is connected to every other router (fully meshed), the updates come directly from the source, which removes the need for a router to propagate any updates that it has received.

9: How is a fully meshed network avoided in iBGP?

** A fully meshed network can cause some problems to the routers because they have a great deal of information to handle from every directly connected router. The configuration of route reflectors does not use as much bandwidth, CPU, or memory. Route reflectors allow the router to pass routes on to its peers, and a hub-and-spoke topology is thus possible.

10: What is the equation to determine the number of sessions needed in a fully meshed BGP network?

** The equation for determining the number of sessions required is n (n – 1) / 2, where n is the number of routers. Thus, 10 routers would mean 10 (10 – 1) / 2 = 10 * 9 / 2 = 45 sessions.

0 Comments:

Post a Comment

Subscribe to Post Comments [Atom]

<< Home