Networking

Frame relay: The straightforward approach

Corporate headquarters has told you to connect the New York City and Los Angeles offices in 24 hours. Panic? No need to when you have frame relay. In this Daily Drill Down, Todd Lammle shows you how to get your bicoastal offices together in no time flat.


In this Daily Drill Down, I’ll discuss a straightforward approach to configuring frame relay using a Cisco frame relay switch connecting three Cisco routers to a corporate office.

To really understand frame relay, you might want to look into what makes up frame relay, as well as some of its history. A good source for that information is the TechProGuild piece “Getting fundamental with frame relay.” This Daily Drill Down discusses the technical aspects of frame relay and how a DLCI, LMI type, and frame relay encapsulation are used in a frame relay network.

Configuring frame relay: The basics
Frame relay is fairly simple to configure. Remember that by default, Cisco uses High-level Data Link Control (HDLC) as a serial link encapsulation. This is a proprietary encapsulation and can only be used if you’re connecting two Cisco routers with a point-to-point link. Frame relay allows you to connect multiple vendors’ routers because of the frame relay standards created. If you had two offices that were connected with a dedicated line, the configuration of the physical network would look like the one shown in Figure A.

Figure A
This setup uses dedicated (read: paid for) links 24/7—a costly endeavor.


The configuration of each router is very simple and looks something like this:

Corporate:
Router>enable
Router#config t
Router(config)#hostname Corp
Corp(config)#interface serial 0
Corp(config-if)#ip address 10.1.1.1 255.255.255.0
Corp(config-if)#no shutdown
Corp(config-if)#exit
Corp(config)#exit
Corp#


Branch:
Router>enable
Router#config t
Router(config)#hostname Branch
Branch(config)#interface serial 0
Branch(config-if)#ip address 10.1.1.2 255.255.255.0
Branch(config-if)#no shutdown
Branch(config-if)#exit
Branch(config)#exit
Branch#


It basically doesn’t get any easier than in this example. The two routers can now communicate point to point. If a company had large streams of data that constantly needed to be sent, this would be a good solution. For the sake of our example, let’s say that this company only sends a few text documents and e-mail. Frame relay would be a good solution for this company because it can provide the same connection speed but at about half the price. The physical network would now look like the one shown in Figure B.

Figure B
We’re no longer required to use Cisco-only routers.


The configuration is only slightly different than the point-to-point configuration:

Corporate:
Router>enable
Router#config t
Router(config)#hostname Corp
Corp(config)#interface serial 0
Corp(config-if)#encapsulation frame-relay
Corp(config-if)#ip address 10.1.1.1 255.255.255.0
Corp(config-if)#frame-relay interface-dlci 100
Corp(config-if)#frame-relay lmi-type cisco
Corp(config-if)#no shutdown
Corp(config-if)#exit
Corp(config)#exit
Corp#


Branch:
Router>enable
Router#config t
Router(config)#hostname Branch
Branch(config)#interface serial 0
Branch(config-if)#encapsulation frame-relay
Branch(config-if)#ip address 10.1.1.2 255.255.255.0
Branch(config-if)#frame-relay interface-dlci 101
Branch(config-if)#frame-relay lmi-type cisco
Branch(config-if)#no shutdown
Branch(config-if)#exit
Branch(config)#exit
Branch#


The encapsulation was changed to frame relay from the default of HDLC and since no encapsulation method was specifically configured, the default is Cisco, which means that a Cisco router is on both sides of the link.

If you have a Cisco router on one side of the link and a Nortel or Juniper router on the other side of the link, for example, then you would use this command:
Branch(config-if)#encapsulation frame-relay ietf

Also, in the configuration example above, the DLCI number for each permanent virtual circuit (PVC) was assigned to each interface (this number is provided to you by the ISP) as well as the LMI type.

The LMI type of Cisco was used because the frame relay switch is a Cisco frame relay switch. (This is not typical. Usually, the Ansi LMI type is used, which means the frame relay switch is anything other than a Cisco.) If you have Cisco routers and a Cisco frame relay switch, you don’t even need to set the LMI type; however, I did it here to show you how to set the LMI type in a configuration. Cisco routers also can auto-detect the LMI type, so if you don’t know what type of switch you’re connecting to, you can let the routers auto-sense the LMI type.

The typical frame relay corporate network
This is where frame relay really shines. If you have multiple locations to connect and you don’t need the dedicated lines to each location, then frame relay is the savior of the WAN world.

Figure C
Notice that each location has a direct physical connection to the frame relay network.


The network would be physically connected, as shown in Figure C.

This setup provides a huge cost savings because you only pay for the toll charges to connect from each location to the nearest provider’s point-of-presence (POP). You then purchase PVCs, which are identified by a Data Link Connection Identifier (DLCI), for each connection you need. For example, the corporate office would purchase three PVCs, one for each connection to a remote branch. The provider would then give the corporate IS department three DLCI numbers to configure under the serial interface of the router, with each DLCI number identifying a connection to a certain location. The brunt of the work is then done by the provider, which maps each and every DLCI to each of the remote locations.

Configuring the corporate router
The recommended configuration for this network is subinterfaces. Figure D shows how the physical connection would look if you use subinterfaces in your frame relay network.

Figure D
The remote routers think they’re connected point to point, but they’re actually connected with a virtual point-to-point connection.


This setup allows the same benefits of dedicated point-to-point links without the cost. You do have a single point of failure, but the idea here is to save money.

Using the original network configuration of three remote branches connected to the corporate office through a frame relay network, the configuration of the corporate router would now look like this:

Corporate:
Router>enable
Router#config t
Router(config)#hostname Corp

Corp(config)#interface serial 0
Corp(config-if)#encapsulation frame-relay
Corp(config-if)#frame-relay lmi-type cisco
Corp(config-if)#no shutdown
Corp(config-if)#no ip address

Corp(config-if)#interface serial 0.20
Corp(config-subif)# ip address 10.20.1.1 255.255.255.0
Corp(config-subif)# frame-relay interface-dlci 20

Corp(config-if)#interface serial 0.21
Corp(config-subif)# ip address 10.21.1.1 255.255.255.0
Corp(config-subif)# frame-relay interface-dlci 21

Corp(config-if)#interface serial 0.22
Corp(config-subif)# ip address 10.22.1.1 255.255.255.0
Corp(config-subif)# frame-relay interface-dlci 22

Corp(config-if)#exit
Corp(config)#exit
Corp#


I broke up the configuration into multiple parts so you can see the configuration better. The first section just sets the hostname of the router; the second section configures the serial 0 interface with the encapsulation LMI type and enables the interface with the no shutdown command. Notice the no ip address command under the serial 0? This is important.

Where to add IP addresses?
If you add an IP address directly under the serial interface, the rest of the configuration will not work.

The next three parts create three subinterfaces, each using a different subinterface number. It doesn’t matter which subinterface number you use. I like to use the assigned DLCI number to help me identify the connection configuration information, but that is for administration purposes only and it matters not what the number is in the actual configuration.

I used interface serial0.20 for PVC connection 20. Notice that under that subinterface configuration, the IP address and DLCI number are identified. I created different IP subnets for each subinterface. I used the DLCI number as the subnet for ease of administration. The remote location for PVC 20 would be IP address 10.20.1.2, the IP address for PVC 21’s remote router would be 10.21.1.2, etc. Instead of using frame relay mapping, I’m allowing IARP to find each PVC and translate the Network layer address to a DLCI number for each connection.

Understand that each configuration under a subinterface must be a separate subnet. That’s the beauty of this configuration. You have more control over how each PVC is configured when you use subinterfaces. For example, if you wanted to add IPX support to the N.Y. office, you could just add an IPX address under the subinterface for the PVC to N.Y. and you’d be set. The PVCs for L.A. and Chicago would not receive IPX packets.

This is a completely different solution than the first frame relay configuration example I showed you in which all devices were in the same IP subnet, and would also all be in the same IPX network, or even the same AppleTalk network, if you decided to use AppleTalk.

The second way to configure the corporate router is the old way and is no longer recommended, but since it still can be done, I’ll give you a configuration example.

In Figure C, the corporate office was provided a PVC DLCI number for each location; however, notice that each of the remote locations uses a different DLCI number to identify itself to the frame relay switch. The DLCI numbers are irrelevant and only locally significant. This means that the corporate office needs a DLCI number to identify each of the remote locations, but this only identifies the transmitted data frame to the switch.

When the frame relay switch receives a data frame from the corporate router with a destination DLCI number of 21, it knows to packet-switch the data to the L.A. PVC. It doesn’t matter whatsoever what DLCI number is used by the L.A. router to communicate to the frame relay switch. Just understand that when you see a DLCI number, it’s the number used by the local router to get to the frame relay switch, which in turn uses the number to figure out which PVC to send the data.

So here’s the first configuration we can use on the corporate router:

Corporate:
Router>enable
Router#config t
Router(config)#hostname Corp
Corp(config)#interface serial 0
Corp(config-if)#encapsulation frame-relay
Corp(config-if)#ip address 10.1.1.1 255.255.255.0
Corp(config-if)#frame-relay interface-dlci 20
Corp(config-if)#frame-relay interface-dlci 21
Corp(config-if)#frame-relay interface-dlci 22
Corp(config-if)#frame-relay map ip 10.1.1.2 20
Corp(config-if)#frame-relay map ip 10.1.1.3 21
Corp(config-if)#frame-relay map ip 10.1.1.4 22
Corp(config-if)#frame-relay lmi-type cisco
Corp(config-if)#no shutdown
Corp(config-if)#exit
Corp(config)#exit
Corp#


The first part is basically the same: You set the encapsulation on the serial interface, add the IP address (if using IP), and then set the DLCI number.

However, you have three DLCI numbers to add, so there are three lines of DLCI addresses.

The new part is the address mapping. Basically, this just says that if you’re sending a packet to 10.1.1.2, for example, use frame relay DLCI number 20. When the frame relay switch receives this frame, it will then forward it through the frame relay packet-switching network to the N.Y. PVC. You don’t have to add the static mappings; you can allow what is called Inverse ARP (IARP) to find each PVC (on by default), which will provide a Network layer-to-Data Link layer translation. You can see these translations with the show frame-relay map command.

Conclusion
You can learn much more about frame relay but, other than a few verification commands, you can pretty much connect to and configure a small frame relay network using the information I provided in this Daily Drill Down.

Just remember the encapsulation types, LMI types and what they do, and how to set DLCIs using subinterfaces and you’re probably going to be up and running with no problem. With frame relay, most of the work is done at the provider’s location and you only have these few commands I discussed to configure.

Editor's Picks

Free Newsletters, In your Inbox