RIP Support

Netcom II supports Routing Information Protocol (RIP) over all LAN and WAN interfaces. To fully support RIP, both RFC1058 (standard RIP over LAN interfaces) and RFC1582 (Triggered RIP over WAN interfaces) were implemented in the /etc/xroute daemon. In a typical system, only one application can supply routing information. If the -r option is used on xroute, then routed must be disabled. For RIP to be supported over SVC or LAPB links, xroute -r must be used.

RIP broadcasts are normally sent over UDP socket 520.

xroute RIP is supported on SCO and SV4 operating systems. To enable RIP broadcasts, xroute must be invoked with the -r option in the shell script /etc/xinet.rc.

Although the actual data transmitted is very similar, the mechanism by which RIP data is sent over the LAN and WAN differ significantly. RIP over a LAN assumes that all systems are always connected and that sending data is fast and inexpensive. For these reasons, RIP data is broadcasted every 30 seconds or whenever a change in the routing tables justifies sending out the RIP data earlier. For a WAN it is unlikely, and usually impossible, to be connected to all hosts at all times. WAN data speeds are usually slow and WAN connections are usually billed by time or amount of data transmitted. For these reasons, RIP data is sent out only during startup and whenever a change in routing tables justifies it. RIP data is also transmitted whenever a remote host requests it. Since WANs do not support broadcast addresses, a table of hosts that support WAN-RIP must be created. This table is found in /etc/x25hosts. Hosts that support WAN-RIP are defined with the RIP flag.

During startup, a connection is made to as many RIP-flagged hosts as possible (based on the number of available SVCs). Each host is requested to send a copy of their routing tables. Once all data to a host is transferred, the normal Inactivity Timer is started, and the virtual circuit to the host is cleared. As VCs are cleared, new VCs can be established to hosts that have not yet been given the routing information. This process continues until all hosts have had their routing tables transferred. At this point the routing tables have stabilized and no further VCs need to be established.

Note: To reduce VC usage, any VCs that are established during normal TCP/IP communication can also be used to transfer RIP data.

Normally, a technique called Split Horizon is used both by LAN and WAN hosts. Split Horizon means that when routing tables are transferred, only desirable routing entries are sent. A desirable element is an entry not learned from the interface to which the routing information is being sent. This helps reduce the size of the routing tables and protects hosts from incorrectly assuming the ability to reach another host through neighboring hosts. If necessary, Split Horizon can be suppressed for individual hosts.

Here are a few example entries that would be placed in /etc/x25hosts.

 
192522301 test1 link00 IP RIP
192522302 test2 link00 IP RIP
192522303 test3 link00 IP
192522304 test4 link01 IP RIP:NOSPLITHORIZON
1         test5 link02 IP HDLC:P

In this example, the test1 and test2 systems are connected to the same X.25 link, support RIP, and use Split Horizon. test3 is also connected to the same link but does not support RIP. test4 is connected to a different X.25 link and supports RIP without using Split Horizon. test5 is connected to a permanent LAPB link and supports RIP.

Revision 6.1.0 (February 2003)