BGP stores the route information it receives from update messages in the routing table, and the routing table exports active routes from the routing table into BGP. BGP then advertises the exported routes to its peers. By default, the exchange of route information between BGP and the routing table occurs immediately after the routes are received. This immediate exchange of route information might cause instabilities in the network reachability information. To guard against this, you can delay the time between when BGP and the routing table exchange route information.
To configure how often BGP and the routing table exchange route information, include the out-delay statement:
-
out-delay seconds;
By default, the routing table retains some of the route information learned from BGP. To have the routing table retain all or none of this information, include the keep statement:
-
keep (all | none);
For a list of hierarchy levels at which you can configure these statements, see the statement summary sections for these statements.
The routing table can retain the route information learned from BGP in one of the following ways:
In an AS path healing situation, routes with looped paths theoretically could become usable during a soft reconfiguration when the AS path loop limit is changed. However, there is a significant memory usage difference between the default and keep all because it is common for a peer to readvertise routes back to the peer from which it learned them. The default behavior is not to waste memory on such routes.