Frame-Relay 101

Frame Relay is a streamlined subset of the X.25 packet switching protocol which has been used by many corporations for wide area communications for a number of years. By removing a number of the X.25 protocol's seldom-used functions and their associated overhead, the Frame Relay protocol allows communications at up to T1 speeds (about 1.5 megabits per second).

The generic advantage provided by Frame Relay is its ability to combine multiple streams of "bursty" data (such as LAN protocol traffic) all of which have relatively low average usage rates, into a single channel with a relatively higher average usage rate. This "statistical multiplexing" effect allows your Frame Relay carrier to provide high bandwidth wide area connectivity to you at a price which is often significantly lower than standard leased line rates.

Virtual Circuits

Like X.25, Frame Relay is a connection oriented service requiring circuits to be configured by your carrier to establish a physical link between two or more locations. Multiple virtual circuits (which appear as virtual point-to-point links) can be run through the same physical connection.

There are two types of virtual circuits supported in Frame Relay: Permanent Virtual Circuits (PVC) and Switched Virtual Circuits (SVC).

PVCs are like dedicated point-to-point private lines. Since the physical connection is always there in the form of a leased line, call setup and tear down is done by a carrier via a network management system.

SVCs are analogous to X.25 connections, which require call setup and tear down.

Note: SVCs are generally not yet available from Frame Relay carriers. Virtually all Frame Relay communications is presently being done using PVCs.

Addressing

A number called the Data Link Connection Identifier (DLCI) identifies each virtual circuit within a shared physical channel.

Frame relay packets are exchanged between nodes by mapping packets containing the source node's DLCI address to the destination DLCI address at the switch. Each switch contains a table identifying the various DLCIs with their associated user lines and interface trunks. However, the switch has more or less work depending on if the DLCI has global or local significance.

Local & Global DLCIs

Local DLCI addressing means that DLCI numbers are only significant at one end of a Frame Relay virtual circuit (VC). In other words, the same VC will be identified by different DLCI's at each end. To accomplish this, a mapping occurs across a VC. Frame Relay switches are required to translate the "source" DLCI in a packet to the "destination" DLCI when forwarding the packet.

Global DLCI addressing is a Local Management Interface (LMI) extension that allows a DLCI number to have universal significance. A global DLCI identifies the same VC at both ends. Global addressing simplifies address administration but allows for only 1024 DLCIs in the entire network. The switch is not required to translate the DLCI in a packet as it does with local DLCI's.

Note: The majority of Frame Relay connections use Local DLCI addressing, where a DLCI number is only significant at one end of the PVC.

Local Management Interface (LMI)

The local management interface specifies communication between different Frame Relay devices (i.e. frame relay switches, routers, access devices, etc.). Over the years, three different local management interface specifications have been developed for Frame Relay: "consortium" LMI (an early cooperative effort by a group of frame relay vendors), CCITT Annex A, and ANSI Annex D. The CCITT and ANSI specifications are formal outgrowths of the consortium LMI specification.

The Annex D specification is the most widely used in the United States, although consortium LMI is still in use by some carriers. The Annex A specification is primarily a European specification.

Encapsulation and Fragmentation

RFC 1490 describes an encapsulation method for carrying packets across a Frame Relay network. All protocol packets are encapsulated within a Q.922 Annex A frame (a CCITT specification for data frames). Additionally, the frames must contain information necessary to identify the protocol being carried, allowing the receiver to properly process the incoming packet.

RFC 1490 also specifies a simple fragmentation procedure for carrying large frames over a frame relay network with a smaller maximum frame size.

Network/Protocol Addressing and Virtual Ports

Routing between LANs across a Frame Relay network is similar to routing across a point-to-point connection. A PVC on one router is directly connected to a PVC on another router. The difference is that multiple PVCs can be supported on the same physical interface of a router.

Network/protocol addresses are associated with each PVC using one of two methods: static mapping, or the Inverse Address Resolution Protocol (IARP).

IARP is outlined in RFC 1293. IARP allows dynamic mapping of protocol addresses to a DLCI. It can be used for IP, IPX and AppleTalk. It is more flexible and easier to configure than static configuration.

IARP is used when a router discovers a new PVC with its corresponding DLCI on a physical interface. The PVC is discovered by communicating with the Frame Relay switch using the LMI protocol. This may be done when the router is coming up or when a PVC has come back up after going down for some reason.