A high-speed packet switching protocol used in wide area networks (WANs). Providing a granular service of up to DS3 speed (45 Mbps), it has become popular for LAN to LAN connections across remote distances, and services are offered by most major carriers.Frame relay (FR) is much faster than X.25, the first packet-switched WAN standard, because frame relay was designed for reliable circuits and performs less error detection (X.25 was never widely used in the U.S.). Frame relay does not process the packets; it relays them from the switch's input port to the output port, hence the name.The FRAD (Frame Relay Access Device)Attachment to a frame relay network is made via a FRAD on the customer's premises, which may be a separate device or software in the router. The FRAD connects to a switch port on the service provider's network via the User-to-Network Interface (UNI). All traffic for one customer generally travels over the same line, which is typically a multiple of 64 Kbps. Frame relay switches interconnect via point-to-point lines or an ATM backbone.Permanent and Switched CircuitsFrame relay provides Permanent Virtual Circuits (PVCs) and Switched Virtual Circuits (SVCs). They are logical connections provisioned ahead of time (PVCs) or on demand (SVCs).Connections are identified by a Data Link Connection Identifier (DLCI) number that is significant only to the local FR switch, which changes the number as it passes the packet on to its destination. The receiving switch uses a different DLCI for its end of the same connection. Every DLCI requires a Committed Information Rate (CIR), which is a pledge on the part of the network to provide a certain amount of transmission capacity for the connection. CIRs are adjusted with experience.Voice Over FRVoice can be packetized to travel over a frame relay network, often providing significant cost savings with some sacrifice in voice quality, depending on network configuration. In 1998, the Frame Relay Forum finalized the Voice Over FR specification. FRF.11 defines the formats, and FRF.12 subdivides large frames in order to interleave real-time voice with data on slow connections.A Superb Resource"Frame Relay for High-Speed Networks" by Walter Goralski is must reading not only to learn about frame relay, but about wide area networking in general. Goralski factors in history, trends and related networking technologies. Published by John Wiley & Sons, Inc., 1999)
The ITU-T I.122 Recommendation (1988), Framework for Providing Additional Packet Mode Bearer Services established the basic framework for a packet communications mode over an integrated services digital network (ISDN). The access protocol specified in ITU-T Q.922 (1992) is Link Access Procedure for Frame Mode Services (LAPF), which is an adaptation of the Link Access Procedure Data channel (LAPD) signaling protocol developed for ISDN. Frame relay was originally intended as an ISDN framing convention for a bearer service, i.e., information-bearing service, anticipated for the ISDN D channel.The D channel is intended primarily for signaling and control purposes, in support of Signaling System 7 (SS7).The D channel runs at 16 kbps for ISDN basic rate interface (BRI) and at 64 kbps for primary rate interface (PRI). Certainly during the early stages of frame relay development, 16 kbps was not a particularly limiting signaling rate, particularly in the context of X.25 packet switching, which often was limited to 9.6 kbps. Over time, however, it became clear that ISDN was far too slow for data communications, even at B channel rates of 64 kbps, and certainly at D channel rates of 16 kbps. So, frame relay became a distinct service, independent of ISDN. In the context of the OSI Reference Model, frame relay standards address the Physical Layer and Data Link Layer, and do not specify internal network operations. Frame relay is analogous to a streamlined and supercharged version of X.25.Although both are designed to support bursty data traffic, frame relay is intended specifically for LAN-to-LAN traffic, but also is used in support of SDLC and many other legacy protocols.Access to a frame relay network is generally over a dedicated digital circuit in the form of a DDS, a Fractional T1, an E-1 (2.048 Mbps), or a T1 (1.544 Mbps). Access via E-3 (34 Mbps) or T3 (45 Mbps) circuits is also generally available. Frame relay statistically multiplexes frames of data over virtual circuits (VCs), with specifications providing for both permanent virtual circuits (PVCs) and switched virtual circuits (SVCs). (Note: SVCs are virtually non-existent because of their additional complexity and cost and the fear of carriers that such a service would cannibalize the PSTN.) The user interface is in a frame relay access device (FRAD) that can be implemented on the customer premises and is analogous to an X.25 packet assembler/disassembler (PAD). Like X.25, frame relay is intended for bursty data traffic, although it works well with fixed bit rate applications, for which it offers assured bandwidth. Although both X.25 and frame relay can support voice, video, and audio, the inherently unpredictable levels of latency and loss over such a highly shared network translate into quality of service (QoS) issues. As frame relay specifies a completely digital network, error performance is excellent. Therefore, frame relay does not attempt to correct any errors created in transit, but simply discards errored frames. It is the responsibility of the receiving user equipment to discover and recover from such an action.As frame relay guarantees frame delivery in the order sent, there is no frame sequence numbering, and there are no acknowledgements of any sort provided.As a result, the load on the computational and bandwidth resources of the network is reduced, frame processing and forwarding are speeded up considerably, and latency is reduced significantly.There are, however, a number of congestion control mechanisms that variously work to provide some assurances of acceptable performance.Whereas frame relay specifies a variable size payload up to 4,096 octets, the Frame Relay Forum (now MFA Forum) developed an Implementation Agreement (IA) that sets the maximum size at 1,600 octets for purposes of interconnectivity and interoperability.This frame size easily supports the largest standard 802.3 Ethernet frame of 1,518 octets. See also B channel, BRI, Data Link Layer, D channel, DDS, E-1, E-3, Fractional T1, FRAD, frame, IA, ISDN, ITU-T, latency, LAPD, LAPF, MFA Forum, OSI Reference Model, packet switch, PAD, Physical Layer, PRI, PVC, QoS, SDLC, signaling and control, signaling rate, SS7, statistical time division mulltiplex, SVC, T1, T3, and X.25.