background image
404 Chapter 6: Routing
network. SAP is used by servers to propagate information that describes their services. CCNAs
are expected to be very familiar with SAP and the routers' roles in forwarding SAP information.
The SAP process works very much like the process used by a distance vector routing protocol.
In fact, SAP uses a concept similar to split horizon to stop a node from advertising SAP
information it learned on an interface with updates sent out that same interface. Each server
sends SAP updates by default every 60 seconds that include the IPX address, server name, and
service type. Every other server and router listens for these updates but does not forward the
SAP packet(s). Instead, the SAP information is added to a SAP table in the server or router; then
the packets are discarded. When that router or server's SAP timer expires, new SAP broadcasts
are sent. As with IPX RIP for routing information, IPX SAP propagates service information
until all servers and routers have learned about all servers.
Client initialization flows provide some insight into why routers need to learn SAP information.
Consider Figure 6-14, which includes the use of the Get Nearest Server (GNS) request and
shows a typical startup with a client configured with a preferred server of Server 2.
The overall goal of Client 1 is to log in to its preferred server, Server 2. The first step is to
connect to some server that has a full SAP table so that the client can learn the IPX address of
its preferred server. (The preferred server name is configured on the client, not the IPX address
of the preferred server.) The router might know the preferred server's name and IPX address in
its SAP table, but no IPX message defined allows the client to query the router for name
resolution. However, an IPX broadcast message asking for any nearby server is defined by IPX:
the GNS request. The router can supply the IPX address of some nearby server (Step 2, in
Figure 6-14) because the router has a SAP table.
Next, the client needs to learn which router to use to forward packets to the server discovered
by its GNS request. RIP requests and replies are used by the client to learn the route from any
router (or server) on the same LAN, as seen in Steps 3 and 4 in Figure 6-14. As a result,
Client 1 knows to use the LA router to deliver packets to network 1001.
After connecting to Server 1, the client learns the IPX address of Server 2, its preferred server
(Steps 5 and 6, in Figure 6-14). The client needs to know the best route to the preferred server's
network; therefore, a RIP request and reply to learn the best next-hop router to network 1002 is
shown in Steps 7 and 8, in Figure 6-14. Finally, packets are sent between the client and Server
2 so that the client can log in; the intervening routers are simply routing the packets.
IPX clients create their own IPX address using the network number in the source address field
of the GNS reply. The GNS reply is always sent by a router or server on the same network as
the client. The client examines the source IPX address of the GNS reply to learn its own IPX
network number. The complete client IPX address is formed by putting that network number
with the MAC address of the client's LAN interface.
ch06.fm Page 404 Monday, March 20, 2000 5:11 PM