Frame Relay

Frame Relay; What Is It and How Does It Work?


Frame Relay is a protocol for digital transmission between small networks, known as local area networks (LANs).  Each network is connected through a frame relay switch to a central location, usually a phone or Internet service provider.  The typical setup consists of several routers connected to ports on a multiplexing device, such as a switch, using traditional services such as 56K or T-carrier lines.  Most Frame Relay networks are public-carrier networks, but it can be used by private enterprise as well.  In the latter case, all equipment is owned by the enterprise and the enterprise is responsible for maintenance.  Frame Relay is the preferred method of connection in wide-area networks because of its speed and reliability. 
In networks that use TCP/IP, data is separated and encapsulated into packets for transfer.  Frame Relay is a data-link level protocol, so it cannot transfer packets in this manner.  Instead, packets are transferred as frames.  Frame Relay does not offer the error correction services of higher level protocols, but simply forwards the frame to the destination device, or endpoint.  Frame headers include a frame check sequence that is used by endpoints to determine whether frames have been damaged during transfer or not.  If the frame arrives damaged, the endpoint discards the frame and requests retransmission.  This handoff of error correction duties to the endpoints, rather than the frame relay switch, makes Frame Relay a very fast method of transfer.
Frame Relay uses permanent virtual circuits to transfer data.  The PVCs are located inside a cloud.  Each network has only one connection to the cloud, but there are multiple possible paths for data frames to take.  The effect of this is that all networks essentially have the same bandwidth available to them.  PVCs provide constant connectivity, so each customer has a continuously-available connection to each network using Frame Relay.  PVCs are identified by endpoints using data-link connection identifiers, abbreviated DLCI.  The DLCI is simply used by the endpoint to identify which connection to use when sending a retransmission request, so assignment is generally left up to the endpoints.  Because DLCIs are assigned locally rather than globally, the same interface may be assigned a different DLCI at each endpoint.  DLCIs can be made global through the use of Local Management Interface.  LMI also allows for the creation of multicast groups.
Although the Frame Relay cloud offers multiple virtual connections to each network, it can still become congested with traffic.  Frame Relay uses congestion notification to avoid congestion.  Each frame header also contains a FECN (forward-explicit congestion notification) bit and a BECN (backward-explicit congestion notification) bit.  The FECN is a value that indicates congestion in the direction from source to destination, while the BECN indicates congestion in the opposite direction.  If a frame experienced congestion during transfer, the switch sets the FECN to 1.  The receiving device will use this information to set the BECN value, and this will be relayed to higher network protocols for processing.  The higher level protocol sets the value of the Discard Eligibility bit.  The DE bit, also contained in the header, tells the switch the order in which a packet may be discarded.  A high DE value means that the frame has a good chance of being discarded, while a low DE value means that the switch will try not to discard the frame.
References:

http://www.pulsewan.com/data101/frame_relay_basics.htm
http://www.networkworld.com/details/497.html?def