IP Routing Protocols

The latest changes are sent when:

The routing database is modified by new data. The latest changes are sent through all interfaces running triggered-on-demand RIP.

RFC-2091 also specifies how packet types are handled in the following manner:

An update request is defined as a request to a peer to send its entire routing database. It is sent:

When the XSR is powered up;

When an interface is brought up.

An update response is defined as a message containing zero or more routes; it is retransmitted at periodic intervals until an update acknowledge is received. It is sent:

In response to an update request. The first response contains no routes. Other update responses will not be sent until an update acknowledge is received. Then the routing database is sent.

At power up. The first update response will contain no routes.

When a port comes up. The first response contains no routes.

When a port is brought down.

When there is fresh routing information to be propagated.

Each update response packet sent to a peer is given a sequence number, a 16-bit unsigned integer.

Responses must be received in order. Updates received with a sequence number out of order is dropped. Packets are accepted if:

A sequence number is one more than the previous;

A sequence number is the same as the previous (occurs when the ack for the previous was sent, but not received on the other side);

The sequence number is 0 (could occur at startup or when it wraps around).

The response sequence number received will be saved and used as a starting point.

Resynchronization occurs with every update response.

Update acknowledgments answer every update response.

The RFC delineates route persistency in the routing database as follows. Entries learned from a triggered response on participating WAN interfaces are permanent, unless certain events occur, in which case entries are marked as unreachable and the hold-down timer started. These events are:

A circuit-down event has been received; all routes learned from that next hop router are marked unreachable.

An update packet with the flush flag set is received; all routes learned from that next hop router are marked unreachable.

Too many retransmissions of an update go unacknowledged. All routes learned from that next hop router are marked unreachable.

An update response for an expired route comes in. That route is marked unreachable. The XSR does not retain alternative routes as they are not needed for the following scenarios:

Dialer and dialer backup connections, which are not both up at the same time. Dialer backup occurs only when the dialer interface goes down (the best route is lost; the back up interface is brought up, then an update request and reply are issued and the new route installed).

XSR User’s Guide 5-13

Page 115
Image 115
Enterasys Networks X-PeditionTM manual IP Routing Protocols