Edited by Anthony Bruno
In addition to IP, Enhanced IGRP (EIGRP) supports two other network-level protocols: AppleTalk and Novell Internetwork Packet Exchange (IPX). Each of these has protocol-specific, value-added functionality. IPX Novell EIGRP supports incremental Service Advertisement Protocol (SAP) updates, removes the Routing Information Protocol (RIP) limitation of 15 hop counts, and provides optimal path use. A router running AppleTalk EIGRP supports partial, bounded routing updates and provides load sharing and optimal path use.
The case studies provided here discuss the benefits and considerations involved in integrating EIGRP into the following types of networks:
Novell IPX—. The existing IPX network is running RIP and SAP.
AppleTalk—. The existing AppleTalk network is running the Routing Table Maintenance Protocol (RTMP).
This case study illustrates the integration of EIGRP into a Novell IPX network in two phases: configuring an IPX network and adding EIGRP to the IPX network. The key considerations for integrating EIGRP into an IPX network running RIP and SAP are as follows:
Route selection
Redistribution and metric handling
Redistribution from IPX RIP to EIGRP and vice versa
Reducing SAP traffic
Cisco's implementation of Novell's IPX protocol provides all the functions of a Novell router. In this case study, routers are configured to run Novell IPX (see Figure 17-1).
The configuration commands to enable IPX routing for Router A are as follows:
ipx routing interface ethernet 0 ipx network 2ad interface ethernet 1 ipx network 3bc
EIGRP for a Novell IPX network has the same fast rerouting and partial update capabilities as EIGRP for IP. In addition, EIGRP has several capabilities designed to facilitate the building of large, robust Novell IPX networks.
The first capability is support for incremental SAP updates. Novell IPX RIP routers send out large RIP and SAP updates every 60 seconds. This can consume substantial amounts of bandwidth. EIGRP for IPX sends out SAP updates only when changes occur and sends only changed information.
The second capability that EIGRP adds to IPX networks is the potential to build large networks. IPX RIP networks have a diameter limit of 15 hops. EIGRP networks can have a diameter of 224 hops.
The third capability that EIGRP for Novell IPX provides is optimal path selection. The RIP metric for route determination is based on ticks with hop count used as a tie-breaker. If more than one route has the same value for the tick metric, the route with the least number of hops is preferred. Instead of ticks and hop count, IPX EIGRP uses a combination of these metrics: delay and bandwidth. For an illustration of how IPX EIGRP provides optimal path selection, see Figure 17-2.
Both Ethernet and FDDI interfaces have a tick value of 1. If configured for Novell RIP, Router A will choose the Ethernet connection via Network 4 to reach Network 5 because Router D is only one hop away from Router A. However, the fastest path to Network 5 is two hops away, via the FDDI rings. With IPX EIGRP configured, Router A will automatically take the optimal path through Routers B and C to reach Network 5.
To add EIGRP to a Novell RIP and SAP network, configure EIGRP on the Cisco router interfaces that connect to other Cisco routers also running EIGRP. Configure RIP and SAP on the interfaces that connect to Novell hosts and to Novell routers that do not support EIGRP.
In Figure 17-3, Routers E, F, and G are running IPX EIGRP. Router E redistributes EIGRP route information via Network AA to Router D.
The configuration for Router E is as follows:
ipx routing interface ethernet 0 ipx network AA interface serial 0 ipx network 20 interface serial 1 ipx network 30 ipx router eigrp 10 network 20 network 30 ipx router rip no network 20
With EIGRP configured, periodic SAP updates are replaced with EIGRP incremental updates when an EIGRP peer is found. Unless RIP is explicitly disabled for an IPX network number, as shown for Network 20, both RIP and EIGRP will be active on the interface associated with that network number. Based on the preceding configuration, and assuming an EIGRP peer on each EIGRP-configured interface, RIP updates are sent on Networks AA and 30, while EIGRP routing updates are sent on Networks 20 and 30. Incremental SAP updates are sent on Network 20 and Network 30, and periodic SAP updates are sent on Network AA.
The configuration for Router F is as follows:
ipx routing interface ethernet 0 ipx network 45 interface serial 0 ipx network 30 ipx router eigrp 10 network 30 network 45
Partial output for a show ipx route command on Router E indicates that Network 45 was discovered using EIGRP (E), whereas network BB was discovered via a RIP (R) update:
R Net 3bc R Net 2ad C Net 20 (HDLC), is directly connected, 66 uses, Serial0 C Net 30 (HDLC), is directly connected, 73 uses, Serial1 E Net 45 [2195456/0] via 30.0000.0c00.c47e, age 0:01:23, 1 uses, Serial1 C Net AA (NOVELL-ETHER), is directly connected, 3 uses, Ethernet0 R Net BB [1/1] via AA.0000.0c03.8b25, 48 sec, 87 uses, Ethernet0
Partial output for a show ipx route command on Router F indicates that Networks 20, AA, and BB were discovered using EIGRP (E):
E Net 20 [2681856/0] via 30.0000.0c01.f0ed, age 0:02:57, 1 uses, Serial0 C Net 30 (HDLC), is directly connected, 47 uses, Serial0 C Net 45 (NOVELL-ETHER), is directly connected, 45 uses, Ethernet0 E Net AA [267008000/0] via 30.0000.0c01.f0ed, age 0:02:57, 1 uses, Serial0 E Net BB [268416000/2] via 30.0000.0c01.f0ed, age 0:02:57, 11 uses, Serial0
A show ipx servers command on Router E shows that server information was learned via periodic (P) SAP updates:
Codes: S - Static, I - Incremental, P - Periodic, H - Holddown 5 Total IPX Servers Table ordering is based on routing and server info Type Name Net Address Port Route Hops Itf P 4 Networkers 100.0000.0000.0001:0666 2/02 2 Et1 P 5 Chicago 100.0000.0000.0001:0234 2/02 2 Et1 P 7 Michigan 100.0000.0000.0001:0123 2/02 2 Et1 P 8 NetTest1 200.0000.0000.0001:0345 2/02 2 Et1 P 8 NetTest 200.0000.0000.0001:0456 2/02 2 Et1
A show ipx servers command on Router F shows that server information was learned via incremental (I) SAP updates allowed with EIGRP:
Codes: S - Static, I - Incremental, P - Periodic, H - Holddown 5 Total IPX Servers Table ordering is based on routing and server info Type Name Net Address Port Route Hops Itf I 4 Networkers 100.0000.0000.0001:0666 268416000/03 3 Se0 I 5 Chicago 100.0000.0000.0001:0234 268416000/03 3 Se0 I 7 Michigan 100.0000.0000.0001:0123 268416000/03 3 Se0 I 8 NetTest1 200.0000.0000.0001:0345 268416000/03 3 Se0 I 8 NetTest 200.0000.0000.0001:0456 268416000/03 3 Se0
A show ipx eigrp topology command on Router E shows that the state of the networks is passive (P) and that each network provides one successor, and it lists the feasible distance (FD) of each successor via a neighbor to the destination. For network 45, for example, the neighbor is located at address 0000.0c00.c47e and the computed/advertised metric for that neighbor to the destination is 2195456/281600:
IPX EIGRP Topology Table for process 10
Codes: P - Passive, A - Active, U - Update, Q - Query, R - Reply,
r - Reply status
P 20, 1 successors, FD is 1
via Connected, Serial0
P 30, 1 successors, FD is 1
via Connected, Serial1
P 45, 1 successors, FD is 2195456
via 30.0000.0c00.c47e (2195456/281600), Serial1
P AA, 1 successors, FD is 266496000
via Redistributed (266496000/0),
P BB, 1 successors, FD is 267904000
via Redistributed (267904000/0),
The output for a show ipx eigrp topology command on Router F lists the following information:
IPX EIGRP Topology Table for process 10
Codes: P - Passive, A - Active, U - Update, Q - Query, R - Reply,
r - Reply status
P 20, 1 successors, FD is 2681856
via 30.0000.0c01.f0ed (2681856/2169856), Serial0
P 30, 1 successors, FD is 1
via Connected, Serial0
P 45, 1 successors, FD is 1
via Connected, Ethernet0
P AA, 1 successors, FD is 267008000
via 30.0000.0c01.f0ed (267008000/266496000), Serial0
P BB, 1 successors, FD is 268416000
via 30.0000.0c01.f0ed (268416000/267904000), Serial0
IPX EIGRP routes are automatically preferred over RIP routes regardless of metrics unless a RIP route has a hop count less than the external hop count carried in the EIGRP update (for example, a server advertising its own internal network).
Redistribution is automatic between RIP and EIGRP, and vice versa. Automatic redistribution can be turned off using the no redistribute command. Redistribution is not automatic between different EIGRP autonomous systems.
The metric handling for integrating RIP into EIGRP is bandwidth plus delay, left shifted by 8 bits. The metric handling for EIGRP to RIP is the external metric plus 1. An IPX EIGRP router that is redistributing RIP into EIGRP takes the RIP metric associated with each RIP route, increments it, and stores that metric in the EIGRP routing table as the external metric.
In Figure 17-4, a Novell IPX server with an internal network number of 100 advertises this network number using RIP on Network 222. Router A hears this advertisement and installs it in its routing table as being one hop and one tick away. Router A then announces this network to Router B on Network 501 using EIGRP.
The configuration for Router A is as follows:
ipx routing ! interface ethernet 0 ipx network 222 ! interface serial 0 ipx network 501 ! ipx router eigrp 9000 network 222 network 501 ! !The following commands turn off IPX RIP on the serial interface: ! ipx router rip no network 501
The configuration for Router B is as follows:
ipx routing ! interface ethernet 0 ipx network 601 ! interface serial 0 ipx network 501 ipx router eigrp 9000 network 501 network 601 ! !The following command turns off IPX RIP on this router: ! no ipx router rip
The configuration for Router C is as follows:
ipx routing ! interface ethernet 0 ipx network 333 ! interface ethernet 1 ipx network 601 ! ipx router eigrp 9000 network 333 network 601 ! !The following commands turn off IPX RIP on ethernet 1: ! ipx router rip no network 601
The configuration for Router D is as follows:
ipx routing ! interface ethernet 0 ipx network 333 ! interface ethernet 1 ipx network AAA
The output from a show ipx route command on Router A is as follows:
R Net 100 [1/1] via 222.0260.8c4c.4f22, 59 sec, 1 uses, Ethernet0 C Net 222 (ARPA), is directly connected, 1252 uses, Ethernet0 E Net 333 [46277376/0] via 501.0000.0c05.84bc, age 0:04:07, 1 uses, Serial0 C Net 501 (HDLC), is directly connected, 3908 uses, Serial0 E Net 601 [46251776/0] via 501.0000.0c05.84bc, age 5:21:38, 1 uses, Serial0 E Net AAA [268441600/2] via 501.0000.0c05.84bc, age 0:16:23, 1 uses, Serial0
The output from a show ipx route command on Router B is as follows:
E Net 100 [268416000/2] via 501.0000.0c05.84b4, age 0:07:30, 2 uses, Serial0 E Net 222 [267008000/0] via 501.0000.0c05.84b4, age 0:07:30, 1 uses, Serial0 E Net 333 [307200/0] via 601.0000.0c05.84d3, age 0:07:30, 1 uses, Ethernet0 C Net 501 (HDLC), is directly connected, 4934 uses, Serial0 C Net 601 (NOVELL-ETHER), is directly connected, 16304 uses, Ethernet0 E Net AAA [267929600/2] via 601.0000.0c05.84d3, age 0:14:40, 1 uses, Ethernet0
The output from a show ipx route command on Router C is as follows:
E Net 100 [268441600/2] via 601.0000.0c05.84bf, age 0:07:33, 1 uses, Ethernet1 E Net 222 [267033600/0] via 601.0000.0c05.84bf, age 0:07:34, 1 uses, Ethernet1 C Net 333 (NOVELL-ETHER), is directly connected, 15121 uses, Ethernet0 E Net 501 [46251776/0] via 601.0000.0c05.84bf, age 0:07:32, 9 uses, Ethernet1 C Net 601 (NOVELL-ETHER), is directly connected, 1346 uses, Ethernet1 R Net AAA [1/1] via 333.0000.0c05.8b25, 35 sec, 1 uses, Ethernet0
The output from a show ipx route command on Router D is as follows:
R Net 100 [8/2] via 333.0000.0c05.84d1, 18 sec, 1 uses, Ethernet0 R Net 222 [6/1] via 333.0000.0c05.84d1, 18 sec, 1 uses, Ethernet0 R Net 333 [1/1] via 333.0000.0c05.84d1, 18 sec, 1 uses, Ethernet0 R Net 501 [3/1] via 333.0000.0c05.84d1, 17 sec, 3 uses, Ethernet0 R Net 601 [1/1] via 333.0000.0c05.84d1, 18 sec, 1 uses, Ethernet0 C Net AAA (SNAP), is directly connected, 20 uses, Ethernet1
The EIGRP metric is created using the RIP ticks for the delay vector. The hop count is incremented and stored as the external metric. The external delay is also stored. Router B computes the metric to Network 100 given the information received from Router A and installs this in its routing table. In this case, the tick value for Network 100 is 8.
The "2" after the slash in the routing entry for Network 100 is the external metric. This number does not increase again while the route is in the EIGRP autonomous system. Router C computes the metric to Network 100 through Router B and stores it in its routing table. Finally, Router C redistributes this information back into RIP with a hop count of 2 (the external metric) and a tick value derived from the original tick value of the RIP route (1) plus the EIGRP delay through the autonomous system converted to ticks.
Novell IPX RIP routers send out large RIP and SAP updates every 60 seconds regardless of whether a change has occurred. These updates can consume a substantial amount of bandwidth. You can reduce SAP update traffic by configuring EIGRP to do incremental SAP updates. When EIGRP is configured for incremental SAP updates, the updates consist only of information that has changed and the updates are sent out only when a change occurs, thus saving bandwidth.
When you configure EIGRP for incremental SAP updates, you can do the following:
Retain RIP, in which case only the reliable transport of EIGRP is used for sending incremental SAP updates. (This is the preferred configuration over bandwidth-sensitive connections.)
Turn off RIP, in which case EIGRP replaces RIP as the routing protocol.
Figure 17-5 shows a bandwidth-sensitive topology in which configuring incremental SAP updates is especially useful. The topology consists of a corporate network that uses a 56-kbps Frame Relay connection to communicate with a remote branch office. The corporate network has several Novell servers, each of which advertises many services. Depending on the number of servers and the number of advertised services, a large portion of the available bandwidth could easily be consumed by SAP updates.
Router A is configured as follows:
ipx routing ! interface ethernet 0 ipx network 100 ! interface serial 0 encapsulation frame-relay ! interface serial 0.1 point-to-point ipx network 200 ipx sap-incremental eigrp 90 rsup-only frame-relay interface-dlci 101 ! ipx router eigrp 90 network 200
The ipx routing global configuration command enables IPX routing on the router.
The ipx network interface configuration command enables IPX routing on Ethernet interface 0 for Network 100.
For serial interface 0, the encapsulation frame-relay interface configuration command establishes Frame Relay encapsulation using Cisco's own encapsulation, which is a 4-byte header, with 2 bytes to identify the DLCI and 2 bytes to identify the packet type.
The interface serial global configuration command establishes a point-to-point subinterface (0.1). Subinterfaces are logical interfaces associated with a physical interface.
Using subinterfaces allows Router A to receive multiple simultaneous connections over a single Frame Relay interface.
The ipx network interface configuration command enables IPX routing on subinterface serial interface 0.1 for Network 200.
The ipx sap-incremental interface configuration command enables the incremental SAP feature. The required eigrp keyword enables EIGRP and its transport mechanism and, in this case, specifies an autonomous system number of 90. Because this command uses the rsup-only keyword, the router sends incremental SAP updates on this link.
The frame-relay interface-dlci interface configuration command associates data link connection identifier (DLCI) 101 with subinterface serial interface 0.1.
The ipx router eigrp global configuration command starts an EIGRP process and assigns to it autonomous system number 90.
The network configuration command enables EIGRP for Network 200.
Router B is configured as follows:
ipx routing ! interface ethernet 0 ipx network 300 ! interface serial 0 encapsulation frame-relay ipx network 200 ipx sap-incremental eigrp 90 rsup-only ! ipx router eigrp 90 network 200
The ipx routing global configuration command enables IPX routing on the router.
The ipx network interface configuration command enables IPX routing on Ethernet interface 0 for Network 300.
On serial interface 0, the encapsulation frame-relay interface configuration command establishes Frame Relay encapsulation using Cisco's own encapsulation, which is a 4-byte header, with 2 bytes to identify the DLCI and 2 bytes to identify the packet type.
The ipx network interface configuration command enables IPX routing on subinterface serial 0 for Network 200.
The ipx sap-incremental interface configuration command enables the incremental SAP feature. The required eigrp keyword enables EIGRP and its transport mechanism and, in this case, specifies an autonomous system number of 90. Because this command uses the rsup-only keyword, the router sends incremental SAP updates on this link.
The ipx router eigrp global configuration command starts an EIGRP process and assigns to it autonomous system number 90.
The network configuration command enables EIGRP for network 200.
This case study illustrates the integration of EIGRP into an existing AppleTalk network in two phases: configuring an AppleTalk network and adding EIGRP to an AppleTalk network. The key considerations for integrating EIGRP into an AppleTalk network are as follows:
Route selection
Metric handling
Redistribution from AppleTalk to EIGRP and vice versa
Cisco routers support AppleTalk Phase 1 and AppleTalk Phase 2. For AppleTalk Phase 2, Cisco routers support both extended and nonextended networks. In this case study, Routers A, B, and C are running AppleTalk, as illustrated in Figure 17-6.
The configuration for Router A is as follows:
appletalk routing interface ethernet 0 appletalk cable-range 10-10 appletalk zone casestudy interface serial 0 appletalk cable-range 50-50 appletalk zone casestudy
To add EIGRP to an AppleTalk network, configure EIGRP on the interface that connects to the routers. Do not disable RTMP on the interfaces that connect to AppleTalk hosts or that connect to AppleTalk routers that do not support EIGRP. RTMP is enabled by default when AppleTalk routing is enabled and when an interface is assigned an AppleTalk cable range.
In this case study, Routers D and E are running AppleTalk EIGRP. Routers F and G run both AppleTalk and AppleTalk EIGRP. Router G redistributes the routes from the AppleTalk network to the AppleTalk EIGRP network, and vice versa (see Figure 17-7).
The configuration for Router G is as follows:
appletalk routing eigrp 1 interface ethernet 1 appletalk cable-range 125-125 appletalk zone Marketing Lab appletalk protocol eigrp interface serial 1 appletalk cable-range 126-126 appletalk zone WAN appletalk protocol eigrp no appletalk protocol rtmp
The configuration for Router F is as follows:
appletalk routing eigrp 2 interface serial 0 appletalk cable-range 126-126 appletalk zone WAN appletalk protocol eigrp no appletalk protocol rtmp
A show appletalk route command on Router G shows that the first set of routes is learned from an RTMP update, that the second set of routes is directly connected, and that the last route is learned by AppleTalk EIGRP via serial interface 1:
R Net 103-103 [1/G] via 125.220, 0 sec, Ethernet1, zone Marketing Lab R Net 104-104 [1/G] via 125.220, 1 sec, Ethernet1, zone Marketing Lab R Net 105-105 [1/G] via 125.220, 1 sec, Ethernet1, zone Marketing Lab R Net 108-108 [1/G] via 125.220, 1 sec, Ethernet1, zone Marketing Lab C Net 125-125 directly connected, Ethernet1, zone Marketing Lab C Net 126-126 directly connected, Serial1, zone Wan E Net 127-127 [1/G] via 126.201, 114 sec, Serial1, zone Networkers
A show appletalk route command on Router F shows that routes are learned from AppleTalk EIGRP:
E Net 103-103 [2/G] via 126.220, 519 sec, Serial0, zone Marketing Lab E Net 104-104 [2/G] via 126.220, 520 sec, Serial0, zone Marketing Lab E Net 105-105 [2/G] via 126.220, 520 sec, Serial0, zone Marketing Lab E Net 108-108 [2/G] via 126.220, 520 sec, Serial0, zone Marketing Lab E Net 125-125 [1/G] via 126.220, 520 sec, Serial0, zone Marketing Lab C Net 126-126 directly connected, Serial0, zone Wan C Net 127-127 directly connected, Ethernet1, zone Networkers
AppleTalk EIGRP routes are automatically preferred over Routing Table Maintenance Protocol (RTMP) routes. Whereas the AppleTalk metric for route determination is based on hop count only, AppleTalk EIGRP uses a combination of these configurable metrics: delay, bandwidth, reliability, and load.
The formula for converting RTMP metrics to AppleTalk EIGRP metrics is hop count multiplied by 252524800. This is a constant based on the bandwidth for a 9.6-kbps serial line and includes an RTMP factor. An RTMP hop distributed into EIGRP appears as a path slightly worse than an EIGRP-native, 9.6-kbps serial link. The formula for converting EIGRP to RTMP is the value of the EIGRP external metric plus 1.
Redistribution between AppleTalk RTMP and EIGRP and vice versa is automatic by default. Redistribution involves converting the EIGRP metric back into an RTMP hop count metric. In reality, there is no conversion of an EIGRP composite metric into a RTMP metric. Because a hop count is carried in an EIGRP metric tuple as the EIGRP route spreads through the network, 1 is added to the hop count, carried in the EIGRP metric blocks through the network, and put into any RTMP routing tuple generated.
There is no conversion of an EIGRP metric back into an RTMP metric because, in reality, what RTMP uses as a metric (the hop count) is carried along the EIGRP metric all the way through the network. This is true of EIGRP-derived routes and routes propagated through the network that were originally derived from an RTMP route.
This case study illustrates the integration of EIGRP in Novell and Appletalk networks. To add EIGRP to IPX networks, it is critical to configure RIP and SAP on interfaces connecting to Novell hosts or routers that do not support EIGRP. When adding EIGRP to AppleTalk networks, turn off RTMP on the interfaces configured to support EIGRP.