By default, Cisco IOS automatically senses the LMI type and automatically discovers the mapping between DLCI and next-hop IP addresses (using Inverse ARP).  If you use all Cisco routers, the default to use Cisco encapsulation works without any additional configuration.  If you design the Frame Relay network to use a single subnet, you can configure the routers to use their physical interfaces without any subinterfaces – making the configuration shorter still.  In fact, using as many default settings as possible, the only new configuration command for Frame Relay, as compared to point-to-point WANs, is the encapsulation frame-relay command.

Planning a Frame Relay Configuration

When planning for new sites, you must consider the following items and communicate them to the Frame Relay provider, which in turn has some impact on the routers’ Frame Relay configurations:

* Define which physical sites need a Frame Relay access link installed, and define the clock rate (access rate) used on each link.

* Define each VC by identifying the endpoints and setting the CIR.

* Agree to an LMI type (usually dictated by the provider).

For these items, the enterprise engineer does not need to consult the Frame Relay provider.

* Choose the IP subnetting scheme: one subnet for all VCs, one subnet for each VC, or a subnet for each fully meshed subset.

* Pick whether to assign the IP addresses to physical, multipoint, or point-to-point subinterfaces.

* Choose which VCs need to use IETF encapsulation instead of the default value of “cisco”.  IETF encapsulation is typically used when one router is not a Cisco router.

The following list summarizes the configuration steps:

Step 1: Configure the physical interface to use Frame Relay encapsulation (encapsulation frame-relay interface subcommand).

Step 2: Configure an IP address on the interface or subinterface (ip address subcommand).

Step 3: (Optional) Manually set the LMI type on each physical serial interface (frame-relay lmi-type interface subcommand).

Step 4: (Optional) Change the default encapsulation of cisco to ietf by doing the following:

a. For all VCs on the interface add the ietf keyword to the encapsulation frame-relay interface subcommand.

b. For a single VC, add the ietf keyword to the frame-relay interface-dlci interface subcommand (point-to-point subinterfaces only) or to the frame-relay map command.

Step 5: (Optional) If you aren’t using the (default) Inverse ARP to map the DLCI to the next-hop router’s IP address, define static mapping using the frame-relay map ip dlci ip-address broadcast subinterface subcommand.

Step 6: On subinterfaces, associate one (point-to-point) or more (multipoint) DLCIs with the subinterface in one of two ways:

a. Using the frame-relay interface-dlci dlci subinterface subcommand.

b. As a side effect of static mapping using the frame relay map ip dlci ip-address broadcast subinterface subcommand.

A Fully Meshed Network with One IP Subnet

The encapsulation frame-relay command tells the routers to use Frame Relay data-link protocols instead of the default, which is HDLC.

IOS default settings:

* The LMI type is automatically sensed.

* The default encapsulation is Cisco instead of IETF.

* PVC DLCIs are learned via LMI status messages.

* Inverse ARP is enabled by default and is triggered when the status message declaring that the VCs are up is received.