Chapter 9. Distance Vector Protocols: Routing Information Protocol Versions 1 and 2 (RIP-1 and RIP-2)

Configuring RFC 2684

The complete theory behind RFC 2684 is covered in Cisco ATM Solutions . Here, I want to remind you that RFC 2684 (formerly RFC 1483) is an encapsulation method of all routed or bridged protocols over ATM. My formula for RFC 2684 is as follows :

RFC 2684 is the multiprotocol encapsulation method. The encapsulation method of multiple protocols (Layer 3 or bridged) over a single VC is accomplished using LLC/SNAP; the encapsulation method of a single protocol over a single VC is done using mux. Both PVCs and SVCs are supported, provided that the ATM cloud can support PVCs or SVCs.

A permanent virtual circuit (PVC) is a statically defined route, whereas a switched virtual circuit (SVC) is a dynamically defined route through signaling. In this section, you learn the principles behind the implementation of both PVCs and SVCs.

RFC 2684 is a basic encapsulation method, without any "magic" behind it. Your protocol addresses must be manually mapped to the ATM addresses by using either the VPI/VCI pair in the PVC implementation or the NSAP/E.164 address in the SVC implementation.

Let's examine the details of PVC and SVC methods of implementations .

PVC Implementation

The PVC implementation is a static and tedious process from within the ATM network itself. If you implement the PVC-based ATM networks, please be careful to make sure that the information on the VPI/VCI numbers is precise. One mistake on a VPI/VCI number(s) can lead to devastating unpredictable results, like ending up in Moscow instead of NewYork for your connection. Recall that VPI/VCI numbers are locally significant addresses used in ATM's PVC implementations.

For the purpose of this chapter discussion, assume that the ATM network has been configured correctly.

You can implement PVC networks by using one of two methods:

• Static VPI/VCI assignment, where you have to enter VPI/VCI numbers manually based on the providers information

• Dynamic VPI/VCI assignment, where your edge routers obtain VPI/VCI numbers dynamically from the edge ATM switch

In either method, VPI/VCI numbers are locally significant.

NOTE

VPI/VCI numbers are similar to DLCIs in Frame Relay networks, where the DLCI is locally significant. The local significance applies all the way to the interface level. That is, if various router ATM physical interfaces have identical VPI/VCI numbers, they are treated as independent connections to the ATM network.

The static configuration involves full manual intervention. If VPI/VCI assignments change within the ATM cloud, you need to adjust the configurations within the routers manually. The dynamic discovery of VPI/VCI numbers is much more appealing because the attached routers learn the information from their respective adjacent switches dynamically. The discovered PVCs and their traffic parameters are configured on the ATM main interface or subinterface that you specify. Your router receives the PVC parameter information by using Interim Local Management Interface (ILMI).

Let's examine the syntax and configuration examples for both methods of PVC-based network implementation. Figure 8-2 illustrates our implementation example.

Figure 8-2. PVC-Based Configuration Example

The three routers (A, B, and C) connected to the ATM cloud handle two protocols ”IP and IPX. From the OSI model Level 3 perspective (network layer), all the routers are interconnected through one single IP and IPX networks ”131.108.168.0/24 and network 100. Each of the routers has another IP and IPX network attached.

Static PVC Implementation

Example 8-1 presents the configurations for Routers A, B, and C. Notice that the subinterfaces are used to interconnect routers over the ATM cloud. You can use major interfaces instead. Use of the subinterfaces, however, positions you with a more scalable configuration.

Also, the example illustrates the use of LLC/SNAP encapsulation (as opposed to mux). LLC/SNAP encapsulation positions your VCs to carry multiple protocols, whereas mux encapsulation carries only one protocol.

NOTE

I can think of only one instance when you would need a single protocol to be carried over a single VC: when you require to do traffic shaping of individual protocols. Assume that you have IP and IPX to carry over ATM network. If you need to define different traffic-shaping parameters to IP and IPX, you would deploy mux encapsulation.

Example 8-1 FC 2684 Static PVC Configuration for Multiprotocol Encapsulation

 Router A (config)#  interface atm 0.1 multipoint   ip address 131.108.168.1 255.255.255.0   ipx network 100   atm pvc 10 0 100 aal5snap   atm pvc 11 0 200 aal5snap   map-group pvc-static-routerA-ip   map-group pvc-static-routerA-ipx   map-list pvc-static-routerA-ip   ip 131.108.168.2 atm-vc 10 broadcast   ip 131.108.168.3 atm-vc 11 broadcast   map-list pvc-static-routerA-ipx   ipx 100.0000.0000.0002 atm-vc 10 broadcast   ipx 100.0000.0000.0003 atm-vc 11 broadcast  _______________________________________________________________________ Router B (config)#  interface atm 0.1 multipoint   ip address 131.108.168.2 255.255.255.0   ipx network 100   atm pvc 10 0 200 aal5snap   atm pvc 11 0 210 aal5snap   map-group pvc-static-routerB-ip   map-group pvc-static-routerB-ipx   map-list pvc-static-routerB-ip   ip 131.108.168.1 atm-vc 10 broadcast   ip 131.108.168.3 atm-vc 11 broadcast   map-list pvc-static-routerB-ipx   ipx 100.0000.0000.0001 atm-vc 10 broadcast   ipx 100.0000.0000.0003 atm-vc 11 broadcast  _______________________________________________________________________ Router C (config)#  interface atm 0.1 multipoint   ipx network 100   ip address 131.108.168.3 255.255.255.0   atm pvc 10 0 300 aal5snap   atm pvc 11 0 310 aal5snap   map-group pvc-static-routerC-ip   map-list pvc-static-routerC-ip   ip 131.108.168.1 atm-vc 10 broadcast   ip 131.108.168.2 atm-vc 11 broadcast   map-list pvc-static-routerC-ipx   ipx 100.0000.0000.0001 atm-vc 10 broadcast   ipx 100.0000.0000.0002 atm-vc 11 broadcast  

The configurations are quite simple, with no magic involved: You need to manually assign the PVCs with an encapsulation method ”in this case, by using aal5snap encapsulation ”and then you need to manually map the corresponding next -hop IP and IPX addresses to the corresponding PVC's VPI/VCI that takes you to the destination address of the corresponding protocol. VPI/VCI numbers must be obtained from an entity managing the ATM cloud. Cisco IOS mapping is done using map-lists that are defined in the global configuration mode and then referenced using map-groups. Please note that the example illustrates the use of two map-lists, each one dedicated for a specific protocol (in this case, IP and IPX). Although you can have one map-list referring to both protocols, I strongly recommend deploying a separate a map-list per protocol, for modularity reasons.

Figure 8-3 illustrates the complete syntax of the command, with explanations provided in Table 8-4. For more in-depth information and further examples of the static PVC implementations by using RFC 2684 encapsulation, refer to Cisco ATM Solutions .

Figure 8-3. Static PVC Command Syntax

Table 8-4. Descriptions of atm pvc Command Arguments

Field	Description
vcd	Virtual circuit descriptor. It uniquely identifies the PVC in the router. It must be unique in the entire router because there is a reference to that number from the global configuration level. The vcd numbers stay in the routers.
vpi	Virtual path identifier. This is a part of the ATM VC address. It must match the carrier-provided VPI for a specific destination and must be unique within a physical interface of the router. [1]
vci	Virtual channel identifier. This is a part of the ATM VC address. It must match the carrier-provided VCI for a specific destination and must be unique within a physical interface of the router [1] .
aal-encap	This is a mandatory parameter that identifies the ATM adaptation layer (AAL) and encapsulation type. Figure 8-2 identifies the various encapsulation types.
midlow	This is optional parameter. It is set for the aal34smds encapsulation only. It is the starting message identifier (MID) number for this PVC. The default is 0. If you set the peak and average (burst is optional) values for aal34smds encapsulation, you also must set the midlow and midhigh values. This option is not available for the ATM port adapter.
midhigh	This is optional parameter. It is set for the aal34smds encapsulation only. It is the ending MID number for this PVC. The default is 0. If you set the peak and average (burst is optional) values for aal34smds encapsulation, you also must set the midlow and midhigh values.
peak	This is an optional parameter that signifies the maximum rate at which this virtual circuit can transmit, in kilobits per second. By default, peak = 155,000 kbps. [2]
average	This is an optional parameter that signifies the average rate at which this virtual circuit transmits, in kilobits per second. By default, average = 155,000 kbps. [2]
burst	This is an optional parameter. It is the value that relates to the maximum number of ATM cells that the VC can transmit to the network at the peak rate of the PVC. The default value varies, based on the model of the router. For example, the 7XXX series default burst value = 94 [2] , [3] .
inarp x	This is an optional parameter to enable Inverse ARP on the PVC (works only for IP, which is RFC 2225, "Classical IP"). The Inverse ARP datagrams are sent every x minutes on this PVC. The default value is 15 minutes.
oam x	This is an optional parameter that configures the transmission of OAM F5 loopback cells every x seconds. The OAM F5 cells verify connectivity on the virtual circuit. The remote end must respond by echoing back such cells.
Information excerpted from Cisco ATM Solutions , Galina Diker Pildush, Cisco Press.

^[1] The reference to the physical interface should not be confused with the subinterface notation. Cisco routers treat the subinterfaces as real physical interfaces, each of which is a separate broadcast domain that can have unique traffic control and policy being definitions, using access lists, for example. The VPI/VCI must be unique within an actual physical interface range, simply because an ingress ATM switch has no idea how to "spell" the word subinterface.

^[2] The peak, average, and the burst numbers are the PCR, SCR, and BT values that are used by the router to perform traffic shaping using the leaky bucket algorithm, which is discussed in Chapter 4, "ATM Traffic and Network Management," of the Cisco ATM Solutions book.

^[3] The default burst value of 94 for the 7 XXX series signifies that the default burst tolerance is 94 tokens. Each token in the 7 XXX series handles 32 cells. In the 4 XXX series, each token handles only one cell . Therefore, in the Cisco literature, you sometimes see a reference to "cells" in the 4 XXX commands and to "tokens" in the 7 XXX commands.

Dynamic PVC Implementation

The dynamic PVC setup is not as manually intensive as the static method. Still, you use PVCs, which are set up permanently through the ATM cloud. The dynamics here involve automatic discovery of the VPI/VCI numbers by using the Integrated Local Management Interface (ILMI). To use ILMI, you need to define the ILMI PVC, by using VCI = 16, per ATM Forum specifications. When you enter the command atm ilmi-pvc-discovery subinterface, the ILMI PVC carries the PVC identifier. Example 8-2 presents more detail on dynamic PVC implementation.

Example 8-2, based on Figure 8-2, illustrates dynamic PVC implementation.

Example 8-2 Dynamic PVC Configuration for Multiprotocol Encapsulation

 Router A (config)#  interface atm 0   atm pvc 1 0 16 ilmi   atm ilmi-pvc-discovery subinterface   interface atm 0.1 multipoint   ip address 131.108.168.1 255.255.255.0   ipx network 100  _______________________________________________________________________ Router B (config)#  interface atm 0   atm pvc 1 0 16 ilmi   atm ilmi-pvc-discovery subinterface   interface atm 0.1 multipoint   ip address 131.108.168.2 255.255.255.0   ipx network 100  _______________________________________________________________________ Router C (config)#  interface atm 0   atm pvc 1 0 16 ilmi   atm ilmi-pvc-discovery subinterface   interface atm 0.1 multipoint   ip address 131.108.168.3 255.255.255.0   ipx network 100  

Please note the specification of the subinterface keyword in the atm ilmi-pvc-discovery command. This enables the discovered PVCs to reside on an ATM subinterface that is specified below. The discovered PVCs are assigned to the subinterface number that matches the VPI number of the discovered PVC. In the example, the discovered PVCs must have a VPI value of 1 to be assigned to the subinterface 0.1.

Notice that static mapping is not defined. When PVC discovery is enabled on an active PVC and the router terminates that PVC, the PVC generates an ATM Inverse ARP request. This allows the PVC to resolve its own network addresses without configuring a static map.

Address mappings learned through Inverse ARP are aged out. However, mappings are refreshed periodically. This period is configurable using the inarp command, which has a default of 15 minutes. It is interesting to note that Inverse ARP initially was available only for IP, following RFC 2225 (Classical IP). Currently, Cisco IOS Software extends the availability of Inverse ARP to the IPX protocol as well. The "Configuring RFC 2225 (Classical IP)" section of the chapter elaborates more on this subject.

NOTE

The fact that dynamically discovered PVCs use ATM Inverse ARP requests is somewhat similar to the Classical IP behavior. Although Classical IP architecture truly does focus on IP protocol only, Cisco extended this capability to the IPX protocol as well.

SVC Implementation

The SVC implementation is much more dynamic and resilient than the PVC implementation. Why? Simply because SVCs are set up on demand, without manual intervention. If traffic needs to get from point A to point Z, signaling sets up the VC dynamically, using either a static or a dynamic ATM routing protocol, through the ATM cloud. After the VC is set up, the traffic can flow through, utilizing the preset path. After the VC is set, all the traffic from source to destination takes the same path. The beauty of an SVC is that you do not have to worry about ATM network availability (provided, of course, that the entire ATM cloud is alive ). If a problem exists with one of the links that is used for a preset VC, a new VC is set up for your traffic dynamically, using Q.2931 signaling. Several methods exist for SVC setup, using static routes or dynamic routing protocols; that discussion is outside the scope of this book, however, and can be found in Cisco ATM Solutions .

Because SVCs are set up dynamically, the addressing scheme is different from that of PVCs. The PVC's address, which is a VPI/VCI combination, is locally significant. The local significance makes total sense ”no protocol at the ingress of the ATM network carries the ATM address of the destination dynamically through the cloud. The SVC address, on the other hand, is globally significant, which also makes total sense. The ATM edge device that needs to pass on the information received through the ATM cloud is responsible for setting up the VC. It does this by using the signaling protocol, Q.2931, which carries the information about the ATM destination and QoS parameters. Path selection within the ATM cloud and number of hops (ATM switches) for this VC setup is immaterial for the IP layer.

The SVC ATM address format consists of 160 bits or 40 hexadecimal numbers, as illustrated in Figure 8-4. If you use a private ATM network, you would use an NSAP-based ATM address. If you use a public ATM network, you would use an E.164-based ATM address. Cisco ATM Solutions discusses the various ATM address formats in depth.

Figure 8-4. SVC ATM Address Format

Although the address seems quite long and tedious to input (it is expressed in hexadecimal notation), do not fear. The ATM Forum foresaw that this would be a problem and allowed the addresses to be learned dynamically using ILMI. All you have to assign is the end station identifier (ESI), consisting of 12 hexadecimal numbers and 2 hexadecimal selector numbers. The prefix is obtained automatically from the immediately attached ATM switch by using ILMI. Various options are available for the ESI assignment ”you can use the MAC address of one of the LAN interfaces of a router, or you can assign an arbitrary address coinciding with the interface's IP address. For example, if the IP address of the ATM interfaces (or subinterface) is 177.10.168.1, you can assign the ESI to be 0177.1016.8100.

Use of ILMI requires you to set up an ILMI PVC by using VCI=16, as identified by the ATM Forum.

Table 8-5 addresses the similarities and differences of SVC implementation and PVC implementation.

Table 8-5. SVC Versus PVC Implementation

Similarities	Differences
Assigning an ATM address to an interface or subinterface	Different form of the ATM addresses: global addresses
Statically mapping Layer 3 next-hop a ddresses to the corresponding ATM addresses (using global ATM addresses)	Need to create PVCs for signaling and, optionally , ILMI

RFC 2684 implementation includes either manual assignment of full ATM address to the ATM edge devices or manual assignment of only the ESI portion of the NSAP address and deployment of the ILMI. Example 8-3, based on Figure 8-1, illustrates the RFC 2684 SVC implementation for the edge device routers A, B, and C. The configurations presented include RFC 2684 handling of IP and IPX protocols. Full NSAP addresses are assigned to routers A, B, and C.

Example 8-3 RFC 2684 SVC Configuration for Multiprotocol Encapsulation

 Router A (config)#  interface atm 0   atm pvc 5 0 5 qsaal   interface atm 0.1   ip address 138.108.168.1 255.255.255.0   atm nsap-address 47.000100010001000100010001.111011101110.00   map-group ip-routerA   map-group ipx-routerA   map-list ip-routerA   ip 131.108.168.2 atm-nsap 47.000200020002000200020002.222022202220.00 broadcast   ip 131.108.168.3 atm-nsap 47.000300030003000300030003.333033303330.00 broadcast   map-list ipx-routerA   ipx 100.0000.0000.0002 atm-nsap 47.000200020002000200020002.222022202220.00   broadcast   ipx 100.0000.0000.0003 atm-nsap 47.000300030003000300030003.333033303330.00   broadcast  ______________________________________________________________________ Router B (config)#  interface atm 0   atm pvc 5 0 5 qsaal   interface atm 0.1   ip address 138.108.168.2 255.255.255.0   atm nsap-address 47.000200020002000200020002.222022202220.00   map-group ip-routerB   map-group ipx-routerB   map-list ip-routerB   ip 131.108.168.1 atm-nsap 47.000100010001000100010001.111011101110.00 broadcast   ip 131.108.168.3 atm-nsap 47.000300030003000300030003.333033303330.00 broadcast   map-list ipx-routerB   ipx 100.0000.0000.0001 atm-nsap 47.000100010001000100010001.111011101110.00   broadcast   ipx 100.0000.0000.0003 atm-nsap 47.000300030003000300030003.333033303330.00   broadcast  _______________________________________________________________________ Router C (config)#  interface atm 0   atm pvc 5 0 5 qsaal   interface atm 0.1   ip address 138.108.168.3 255.255.255.0   atm nsap-address 47.000300030003000300030003.333033303330.00   map-group ip-routerC   map-group ipx-routerC   map-list ip-routerC   ip 131.108.168.1 atm-nsap 47.000100010001000100010001.111011101110.00 broadcast   ip 131.108.168.2 atm-nsap 47.000200020002000200020002.222022202220.00 broadcast   map-list ipx-routerC   ipx 100.0000.0000.0001 atm-nsap 47.000100010001000100010001.111011101110.00   broadcast   ipx 100.0000.0000.0002 atm-nsap 47.000200020002000200020002.222022202220.00   broadcast  

Example 8-4, based on Figure 8-1, illustrates another RFC 2684 SVC implementation for the same edge devices. This example illustrates the ESI portion of the ATM address assigned to the routers' interfaces. ILMI is used to obtain the NSAP prefix portion from the immediately attached ATM switch. Notice that Example 8-3 had only signaling PVCs, whereas Example 8-4 illustrates the use of two PVCs ”signaling and ILMI. The signaling and ILMI PVCs must be assigned to the major interface.

Example 8-4 RFC 2684 SVC Configuration for Multiprotocol Encapsulation

 Router A (config)#  interface atm 0   atm pvc 5 0 5 qsaal   atm pvc 2 0 16 ilmi   interface atm 0.1   ip address 138.108.168.1 255.255.255.0   atm esi-address 111011101110.00   map-group ip-routerA   map-group ipx-routerA   map-list ip-routerA   ip 131.108.168.2 atm-nsap 47.000200020002000200020002.222022202220.00 broadcast   ip 131.108.168.3 atm-nsap 47.000300030003000300030003.333033303330.00 broadcast   map-list ipx-routerA   ipx 100.0000.0000.0002 atm-nsap 47.000200020002000200020002.222022202220.00   broadcast   ipx 100.0000.0000.0003 atm-nsap 47.000300030003000300030003.333033303330.00   broadcast  _______________________________________________________________________ Router B (config)#  interface atm 0   atm pvc 5 0 5 qsaal   atm pvc 2 0 16 ilmi   interface atm 0.1   ip address 138.108.168.2 255.255.255.0   atm esi-address 222022202220.00   map-group ip-routerB   map-group ipx-routerB   map-list ip-routerB   ip 131.108.168.1 atm-nsap 47.000100010001000100010001.111011101110.00 broadcast   ip 131.108.168.3 atm-nsap 47.000300030003000300030003.333033303330.00 broadcast   map-list ipx-routerB   ipx 100.0000.0000.0001 atm-nsap 47.000100010001000100010001.111011101110.00   broadcast   ipx 100.0000.0000.0003 atm-nsap 47.000300030003000300030003.333033303330.00   broadcast  _______________________________________________________________________ Router C (config)#  interface atm 0   atm pvc 5 0 5 qsaal   atm pvc 2 0 16 ilmi   interface atm 0.1   ip address 138.108.168.3 255.255.255.0   atm esi-address 333033303330.00   map-group ip-routerC   map-group ipx-routerC   map-list ip-routerC   ip 131.108.168.1 atm-nsap 47.000100010001000100010001.111011101110.00 broadcast   ip 131.108.168.2 atm-nsap 47.000200020002000200020002.222022202220.00 broadcast   map-list ipx-routerC   ipx 100.0000.0000.0001 atm-nsap 47.000100010001000100010001.111011101110.00   broadcast   ipx 100.0000.0000.0002 atm-nsap 47.000200020002000200020002.222022202220.00   broadcast  

Summarizing, the syntax of the SVC commands for RFC 2684 is as follows:

 Router (config-if)#  atm nsap-address   nsap-address  

or

 Router (config-if)#  atm esi-address   esi  Router (config-if)#  map-group   name  Router (config)#  map-list   name  Router (config-map-list)#  protocol protocol-address   atm-nsap   atm-nsap-address  [  class   class-name  ][  broadcast  ] 

Table 8-6 summarizes all the parameters that are used for configuring router interconnectivity through the SVC-based ATM networks.

Table 8-6. Argument Description of the SVC-Associated Commands

Argument	Description
nsap-address	Source address, specified as 40 hexadecimal digits.
esi	End-station system identifier (esi), specified as 12 hexadecimal digits. To form a full NSAP address, the 26 hexadecimal-digit prefix is learned dynamically from the ingress ATM switch with help of ILMI.
name	The name of the map list that is created at the global configuration mode. The map list name must be referenced from the interface/subinterface mode by using map-group to activate the map list.
protocol	One of the following keywords: ip, ipx, appletalk, decnet, vines, apollo, and so on, depending on the Layer 3 protocol in use.
protocol-address	Destination address that is being mapped to this SVC.
atm-nsap-address	Destination ATM NSAP address.
class-name	A reference to the traffic parameters map class lists. This is optionally available to change the traffic parameter values from their default values. With the help of map-class statements, you can customize the traffic-shaping parameters for various types of traffic.
broadcast	Keyword that is necessary if protocol broadcast traffic, like routing updates, needs to use the ATM interface.
Table 8-5 is excerpted from Cisco ATM Solutions , Galina Diker Pildush, Cisco Press.

Cisco IOS Software extends traffic-shaping capabilities to SVCs with the help of map-class commands:

 Router(config-map-class)#  atm forward-peak-cell-rate-clp0   rate   atm backward-peak-cell-rate-clp0   rate   atm forward-peak-cell-rate-clp1   rate   atm backward-peak-cell-rate-clp1   rate   atm forward-sustainable-cell-rate-clp0   rate   atm backward-sustainable-cell-rate-clp0   rate   atm forward-sustainable-cell-rate-clp1   rate   atm backward-sustainable-cell-rate-clp1   rate   atm forward-max-burst-size-clp0   cell-count   atm backward-max-burst-size-clp0   cell-count   atm forward-max-burst-size-clp1   cell-count   atm backward-max-burst-size-clp1   cell-count  

The map class is referred to from the map list level by using the class command within the map list, as illustrated in Example 8-5.

Example 8-5 Example of Traffic Engineering Deployment in SVC

 Router(config)#  map-class atm contract-svcs  Router(config-map-class)#  atm forward-peak-cell-rate-clp0 56000  Router(config)#  map-list test ip 200.0.0.1 atm-nsap   44.444400000000000000000000.000000000000.00 class contract-svcs